Perovskite solar cells have become a promising candidate for low-cost photovoltaic applications. In the past five years, the power conversion efficiency (PCE) of these devices has increased from 3.8% to 22.1%.^[1–7] The excellent performance is partly related to the effective charge separation at the interface between the electron transport layers and light harvester layers.^[8] Therefore, the properties of electron transport layers are important for the performance of perovskite solar cells.

Several types of materials can be used for fabricating electron transport layers, such as TiO₂,^[9,10] ZnO,^[11,12] SnO₂,^[13,14] and PC₆₁BM.^[15,16] Among these materials, TiO₂ has been widely used in the perovskite solar cells with conventional device structure, owing to its suitable energy band structure and electron mobility.^[17,18] Although mesoporous TiO₂ layers can facilitate the electron collection at the TiO₂/perovskite interfaces, planar perovskite solar cells without mesoporous layers receive much attention for their simple device structure and reduced hysteresis effect.^[19,20]

As electron transport layers have the functions of electron extraction and transport, their transport properties directly affect the performance of perovskite solar cells. The TiO₂ electron transport layers are widely deposited by spray pyrolysis or spin-coating of Ti(OPr)₄ precursor solutions.^[21,22] These solution deposition methods are cost-effective and scalable, but the deposited TiO₂ layers usually have an amorphous structure, which is harmful to the efficient transport of electrons.^[23] The slow electron transport can lead to enhanced recombination, especially for planar devices fabricated by two-step methods.^[24] Therefore, modification methods of TiO₂ electron transport layers have been investigated to improve these planar devices.^[25,26]

TiO₂ nanoparticles have been widely investigated for their photoelectric and catalysis properties,^[27] and the size and crystallization of these materials can be conveniently controlled in large-scale synthesis.^[27] Incorporating TiO₂ nanoparticles into the amorphous TiO₂ layers may provide additional transport channels for electrons in the electron transport layers, and the length of these channels can be easily controlled by the size of nanoparticles. Although TiO₂ nanoparticles have been applied for producing mesoporous layers in perovskite solar cells, these layers work as the scaffold of perovskite layers in the mesoporous devices, which have different functions from the compact TiO₂ layers of planar devices.^[28]

In this work, we develop composite TiO₂ electron transport layers by a mixed spray pyrolysis method. TiO₂ nanoparticles are incorporated into the Ti(OPr)₄ precursor solutions, and the properties of these composite layers are characterized. By optimizing the size of the nanoparticles, the performance of planar perovskite solar cells with different composite TiO₂ layers are investigated.

For producing perovskite solar cells, fluorine-doped tin oxide (FTO) glass substrates were etched by 1064 nm laser and cleaned by acetone, ethanol, and deionized water in ultrasonic bath, sequentially. Then, TiO₂ electron transport layers were deposited by spray pyrolysis with 2.0 mL 0.05 M Ti(OPr)₂(acetylacetonate)₂ in ethanol solution at 450 °C. For incorporating TiO₂ nanoparticles into the TiO₂ layers, 4.0 mg/mL nanoparticles were ultrasonically dispersed in the 0.05 M Ti(OPr)₂(acetylacetonate)₂/ethanol solution. Three types of nanoparticles, with sizes of 25 nm, 60 nm, and 100 nm were employed, which are labeled to 25 nm NPs, 60 nm NPs, and 100 nm NPs. PbI₂ layers were deposited on TiO₂ compact layers by spin-coating at 4000 rpm for 30 s with 462 mg/mL PbI₂ in N,N-dimethylformamide (DMF) solution. After drying at 70 °C for 5 min, 0.4 mL 10 mg/mL CH₃NH₃I in isopropanol solution was sprayed on the top of PbI₂ layers at 80 °C.^[29] Next, the perovskite films were annealed at 100 °C for 30 min, and they were cleaned with isopropanol solution twice. Then, the samples were annealed on a hot plate at 70 °C for an additional 30 min. When cooled to room temperature, hole transport material (HTM) layers were prepared by spin-coating a solution containing 72.3 mg spiro-MeOTAD, 4-tert-butylpyridine, and lithium bis(trifluoromethanesulfonyl)imide solution (520 mg/mL in acetonitrile) in 1 mL chlorobenzene solvent at 4000 rpm for 20 s. Finally, Ag electrodes were deposited on the HTM layers by vacuum thermal evaporation.

Scanning electron microscopy (SEM) images were obtained using a FEI-MAGELLAN 400 scanning electron microscope. X-ray diffraction (XRD) patterns were obtained using a Rigaku D/max-2550 x-ray diffractometer. Photocurrent density–photovoltage characteristics were recorded from 1.15 V to 0 V by a CHI660 electrochemical workstation. The active area of solar cells is 0.15 cm², which was defined by a mask. AM1.5 illumination was provided by a 3A class solar simulator (UHE-16, ScienceTech Inc.), which was calibrated to one sun by a KG5 filtered Si reference solar cell (certificated by VLSI Standards Inc., traceable to National Renewable Energy Laboratory). IPCE spectra were measured in the DC mode by a controlled monochrometer (BOCIC Inc.), and a calibrated Si cell was used as the reference. Photoluminescence (PL) spectra were detected by a CCD detector (PIXIS256BR, Princeton Instruments Inc.), and the excitation wavelength was 532 nm provided by a low-noise solid state laser (MLL-III-532). Impedance spectra were measured under constant illumination condition over a frequency range from 1 MHz to 1 Hz using a CHI660 electrochemical workstation.

Figure 1 shows the SEM morphology of TiO₂ electron transport layers, which are deposited by the precursor solutions without and with different nanoparticles. For the TiO₂ layers deposited by the precursor solution without nanoparticles in Fig. 1(a), the surface preserves the morphology of FTO layers and the TiO₂ layers smoothly cover the surface of FTO layers.^[30] By incorporating 25 nm TiO₂ nanoparticles in the precursor solutions, the surface roughness of TiO₂ layers increases (Fig. 1(b)). The rough surface reflects the existence of these nanoparticles in the TiO₂ layers. For the TiO₂ layers incorporating 60 nm nanoparticles, the morphology of FTO layers is still largely preserved (Fig. 1(c)). However, large particle clusters emerge in the TiO₂ layers incorporating 100 nm TiO₂ nanoparticles (Fig. 1(d)). As shown in Fig. 1, the surface roughness of TiO₂ layers continuously increases with the increase in nanoparticle size.

Fig. 1.

	Figure Option ViewDownloadNew Window
	Fig. 1. Surface morphology of TiO2 layers without (a) and with incorporation of 25 nm (b), 60 nm (c), and 100 nm (d) TiO2 nanoparticles. The scale bar is 200 nm.

After depositing CH₃NH₃PbI₃ layers on these different TiO₂ layers, we measure the surface morphology of the perovskite layers, as shown in Fig. 2. All of these CH₃NH₃PbI₃ layers have compact structure, but the sizes of the CH₃NH₃PbI₃ grains are obviously different. For the CH₃NH₃PbI₃ layers deposited on the TiO₂ compact layers without nanoparticles, the average size of CH₃NH₃PbI₃ grains is approximately 257 nm (Fig. 2(a)). By incorporating 25 nm nanoparticles into the TiO₂ layers, the average size of CH₃NH₃PbI₃ grains increases to 311 nm (Fig. 2(b)). Furthermore, the average sizes of CH₃NH₃PbI₃ grains increase to 371 nm and 381 nm for the TiO₂ layers incorporating 60 nm and 100 nm nanoparticles, respectively (Figs. 2(c) and 2(d)). Therefore, incorporating nanoparticles in the TiO₂ layers can improve the growth of CH₃NH₃PbI₃ grains. Figure 3 shows the XRD patterns of four CH₃NH₃PbI₃ layers. Except for the diffraction peaks of FTO substrates, all the diffraction peaks can be assigned to the tetragonal phase of CH₃NH₃PbI₃.^[31] It is also noted that the intensity of the CH₃NH₃PbI₃ diffraction peaks are different. By incorporating nanoparticles in the TiO₂ layers, the intensity of CH₃NH₃PbI₃ diffraction peaks is increased remarkably and the TiO₂ layers with 60 nm nanoparticles lead to the highest intensity of diffraction peaks. This indicates that incorporating nanoparticles in the TiO₂ layers is beneficial for the crystallization of CH₃NH₃PbI₃ layers and 60 nm is the optimal size of nanoparticles.

Fig. 2.

	Figure Option ViewDownloadNew Window
	Fig. 2. Surface morphology of CH3NH3PbI3 layers on the TiO2 layers without (a) and with incorporation of 25 nm (b), 60 nm (c), and 100 nm (d) nanoparticles. The scale bar is 500 nm.

Fig. 3.

	Figure Option ViewDownloadNew Window
	Fig. 3. (color online) XRD patterns of CH3NH3PbI3 layers on the TiO2 layers without (W/O NPs) and with incorporation of 25 nm (25 nm NPs), 60 nm (60 nm NPs), and 100 nm (100 nm NPs) nanoparticles.

We fabricate planar perovskite solar cells with these CH₃NH₃PbI₃ layers on different TiO₂ electron transport layers, and the solar-to-electric conversion performance of devices is shown in Fig. 4(a). For the TiO₂ layers without incorporating nanoparticles, the corresponding device has a PCE of 9.9% with a short-circuit current density of 16.9 mA/cm², an open-circuit voltage of 0.99 V, and a fill factor (FF) of 0.59. After incorporating 25 nm nanoparticles into the TiO₂ layers, the PCE increases to 12.5%, which comes from the increases in (18.1 mA/cm²) and (1.06 V). By increasing the size of incorporated nanoparticles to 60 nm, a PCE of 16.7% is achieved with of 22.2 mA/cm² and of 1.10 V. However, further increasing the size of incorporated nanoparticles to 100 nm induces a decrease in PCE (15.1%) with a of 20.5 mA/cm² and of 1.08 V. Although efficient mesoporous devices can be easily prepared, planar devices by two-step methods usually suffer from intensive recombination and low reproducibility. These results indicate that incorporating nanoparticles can improve the TiO₂ electron transport layers and promote the performance of planar perovskite solar cells remarkably. It is noted that the optimized size of nanoparticles is 60 nm.

Fig. 4.

	Figure Option ViewDownloadNew Window
	Fig. 4. (color online) Photocurrent density–photovoltage curves (a), statistics of PCEs (b), steady-state power output (c), and IPCE spectra (d) of devices with different TiO2 electron transport layers.

To confirm the superior performance of these composite electron transport layers, we compare the statistics of PCEs for the devices without and with 60 nm nanoparticles. As shown in Fig. 4(b), the TiO₂ layers with 60 nm nanoparticles lead to an average PCE of 13.6%, which is obviously better than those of devices without nanoparticles incorporation (10.1%). Figure 4(c) shows the steady-state power output of devices without and with 60 nm nanoparticles. For these two types of devices, the onset of power output has similar ramping rates. The device without nanoparticles incorporation has relatively low steady-state power output, which also degrades slightly in the 200 s measurement. The incorporation of 60 nm nanoparticles promotes the device power output significantly, and it does not degrade in the 200 s measurement. This means that incorporation of 60 nm nanoparticles is beneficial to the power output and stability of planar devices.

Figure 4(d) shows the IPCE spectra of devices with four different TiO₂ electron transport layers. All of these devices have similar IPCE values in the short-wavelength range (300–450 nm), whereas the IPCE values in the long-wavelength range (450–800 nm) are obviously different. For the devices without the incorporation of nanoparticles, the IPCE values are relatively low in the long-wavelength range, which reflects the poor collection of photogenerated electrons. After incorporating of nanoparticles with 25 nm and 60 nm in size, the IPCE values increase in the long-wavelength range continuously. However, the values of IPCE decrease markedly by further increasing the size of incorporated nanoparticles (100 nm). As the IPCE spectra can be directly related to the of perovskite solar cells, the trend of IPCE spectra is accordance to that of in Fig. 4(a). At the long-wavelength range, the penetrating depth of incident light is longer than that of short-wavelength light. Then the average transport length of photogenerated electrons is relatively long for the long-wavelength light, which leads to the difficulty of electron collection at TiO₂/perovskite interfaces. The improvement in IPCE values in Fig. 4(d) means the incorporation of TiO₂ nanoparticles can promote the electron collection process, which should be related with the electron extraction process at TiO₂/perovskite interfaces and the transport-recombination process in perovskite layers.

PL is a facile tool for evaluating the charge extraction process at TiO₂/perovskite interfaces.^[32,33] We measure the PL spectra of CH₃NH₃PbI₃ layers on different TiO₂ electron transport layers, and the PL spectrum of bare CH₃NH₃PbI₃ layers without TiO₂ layers is measured as the reference. As shown in Fig. 5, these PL peaks are located at approximately 765 nm, and their peak intensities are dependent on the properties of the TiO₂ underlayers. For the TiO₂ layers without incorporation of nanoparticles, the PL intensity of CH₃NH₃PbI₃ layers decrease to 63% of the intensity in bare CH₃NH₃PbI₃ layers. By incorporating 25 nm nanoparticles in the TiO₂ layers, the quenching effect is obviously enhanced to 17% of the PL peaks in bare CH₃NH₃PbI₃ layers. After increasing the size of incorporated nanoparticles to 60 nm, the PL peaks of CH₃NH₃PbI₃ layers can be remarkably quenched to 2.4% of the PL peaks in bare CH₃NH₃PbI₃ layers. As PL emission of 765 nm comes from the direct recombination of photogenerated electrons with holes in perovskite layers, the intensive quenching effect reflects the reduced density of photogenerated electrons in perovskite layers, which comes from the improved charge extraction process at the TiO₂/perovskite interfaces. However, the quenching effect attenuates slightly to 4.9% of the PL peaks in bare CH₃NH₃PbI₃ layers for the incorporation of 100 nm nanoparticles. Therefore, PL analysis reveals that the incorporation of 60 nm nanoparticles can improve the charge extraction process at the TiO₂/perovskite interfaces. A possible reason for this is that nanoparticles provide additional transport channels for electrons, which stimulates the electron extraction process of TiO₂ layers.

Fig. 5.

	Figure Option ViewDownloadNew Window
	Fig. 5. (color online) PL spectra of bare CH3NH3PbI3 layers (reference), and PL spectra of CH3NH3PbI3 layers on TiO2 layers without (W/O NPs) and with incorporation of 25 nm (25 nm NPs), 60 nm (60 nm NPs), and 100 nm (100 nm NPs) nanoparticles.

To investigate the transport-recombination process of perovskite solar cells, we perform the impedance measurement on the devices at 0.7 V bias potential under constant illumination. Figure 6(a) shows the Nyquist plots of devices with different TiO₂ electron transport layers, and an enlargement of the high-frequency arcs is shown in Fig. 6(b). Two arcs can be distinguished in the Nyquist plots, which can be attributed to the charge transfer process at the TiO₂/perovskite interfaces and the recombination process in perovskite layers, respectively.^[34,35] The charge transfer resistance and recombination resistance can be extracted by fitting the impedance data with the equivalent circuit in Fig. 6(a). For the devices without nanoparticle incorporation, is and is . By incorporating 25 nm nanoparticles into the TiO₂ layers, decreases to and increases to . For the devices with the incorporation of 60 nm nanoparticles, is further reduced to and is further promoted to . A low is beneficial to the charge transfer process at the TiO₂/perovskite interface, and a high is related to the low speed of the recombination process. Therefore, the incorporation of 60 nm nanoparticles in the TiO₂ layers can improve the transport-recombination process in planar perovskite solar cells. However, further increasing the size of the incorporated nanoparticles to 100 nm leads to a slight increase in ( ) and decrease in ( ). Therefore, the impedance analysis indicates that the optimized size of incorporated nanoparticles is 60 nm. A low recombination process can guarantee the transport of photogenerated electrons to the TiO₂/perovskite interfaces, which is particularly important for the electrons excited by long-wavelength light.^[36] This result is accordance with the J–V and IPCE measurements of these devices, as shown in Fig. 4.

Fig. 6.

	Figure Option ViewDownloadNew Window
	Fig. 6. (color online) Nyquist plots of devices with different TiO2 electron transport layers (a) and the enlargement of high-frequency arcs (b).

In summary, we develop composite TiO₂ electron transport layers by a mixed spray pyrolysis method, which incorporates TiO₂ nanoparticles into the precursor solutions. We fabricate planar perovskite solar cells with these TiO₂ layers by a two-step method. It is found that the performance of these TiO₂ layers is dependent on the size of incorporated nanoparticles. For the optimized nanoparticle size of 60 nm, the PCE of the corresponding devices reaches 16.7%, which is remarkably higher than that of devices without nanoparticle incorporation (9.9%). PL and impedance measurements indicate that nanoparticles can promote the charge extraction process at the TiO₂/perovskite interfaces and improve the recombination process in perovskite layers. The convenient operation of spray pyrolysis method guarantees a promising potential for industrial application, and this mixed spray pyrolysis also can be used for other functional films.