Investigation on Low-temperature Annealing Process of Solution-processed TiO2 Electron Transport Layer for Flexible Perovskite Solar Cell

Flexible perovskite solar cells (PSCs) have received increasing attention in wearable and portable devices over the past ten years. The low-temperature process of electron transport layer plays a key role in fabricating flexible PSCs. In this paper, we improve the performance of flexible PSCs by controlling the thermodynamic procedure in the low-temperature annealing process of solution-processed TiO2 layers and modulating the precursor concentration of (6,6)-phenyl c61 butyric acid methyl ester (PC61BM) deposited on fluorine-doped tin oxide (FTO)/TiO2 substrate. The results show that slowing down evaporation rate of residual solvent and adopting PC61BM of appropriate precursor concentration are confirmed to be effective methods to improve the performance of flexible PSCs. We also demonstrate carbon electrode-based flexible PSCs. Our work expands the feasibility of low temperature process for the development of flexible perovskite photodetectors and light-emitting diodes, as well as flexible PSCs.

In 2019, based on solution process, Chu et al. studied the influence of treatment time of UV/O 3 process following the annealing process of TiO 2 film on a hotplate on the properties of TiO 2 ETL and discovered that 80 min-UV/O 3 -treating endowed the flexible PSCs with a best PCE of 7.33% owing to UV/O 3 -induced removal of residual organic solvent as well as enhanced hydrophilicity and conductivity of TiO 2 films [19]. In the same year, You et al. investigated the effect of different solvent used to re-disperse TiO 2 nano-sol on the quality of solution-processed TiO 2 ETL prior to the annealing process of TiO 2 film on a hotplate and demonstrated flexible PSCs of better PCE of 15.8% based on N, N-dimethylformamide (DMF) solvent, which has high zeta potential and low surface tension [20].
In this paper, we prepare the nano TiO 2 films with solution process and perform the annealing process on a hotplate at low temperature (≤150 • C). We control the morphology of TiO 2 ETL by adjusting evaporating rate of the organic solvent in the annealing process on a hotplate, which leads to the change of thermodynamic process in the growth and formation of TiO 2 films. Furthermore, we optimize the ETL with PC 61 BM of different precursor solution to reduce the defects on the surface of TiO 2 layer [21]. It is worth noting that all the exploring experiments are carried out on the rigid substrates at first, and then we prepare the carbon electrode-based flexible PSCs based on the optimized preparing process. (1) Giving the heating mode, we determine three kinds of annealing procedures (Process 1): (a) The annealing temperature was directly raised to 150 • C (Direct method); (b) The annealing temperature was raised from room temperature to 150 • C at rate of 8 • C/min (Proportional method); (c) The as-grown nano TiO 2 layer had been delayed for one hour before the annealing temperature was raised from room temperature to 150 • C at rate of 8 • C/min (Delayed method);

Experiments
(2) Based on Process 1(c), we demonstrated an interface modifying procedure (Process 2): Depositing PC 61 BM of different precursor concentration (5, 10, 15, 20, 25 mg/mL) on nano TiO 2 layer by spin-coating at 1500 rpm for 30 s. After the deposition, PC 61 BM was left in the ambient condition for forty minutes to evaporate slowly with no annealing process. It is worth noting that the PC 61 BM was dissolved in chlorobenzene (Shanghai Mater Win New Materials Co., Ltd., Shanghai, China).

Formation of Perovskite Light-absorption Layer
First, the PbI 2 was fully dissolved in a mixed solvent of DMF and DMSO in the volume ratio of 0.95:0.05 by vigorous stirring to form 600 mg/mL precursor solution. The methyl ammonium was fully dissolved in anhydrous isopropanol to form 70 mg/mL precursor solution. Then, the PbI 2 precursor solution was deposited on the PC 61 BM layer by spin-coating at 1500 rpm for 30 s. Immediately, the CH 3 NH 3 I precursor solution was drop-cast on the as-grown PbI 2 film at 1500 rpm for 30 s and then dried at 150 • C for 20 min.

Formation of Hole Transport Layer
The hole transport layer was formed by spin-coating Spiro-OMeTAD on perovskite layer at 3000 rpm for 30 s. After the spin-coating, Spiro-OMeTAD was left in the ambient condition for 40 min to evaporate slowly with no annealing process.

Formation of Counter Electrode
The soot from the burning candle was collected with an FTO glass substrate to form a sponge-like carbon electrode, which was then pressed against the prepared hole transport layer [22].

Characterization
Field emission transmission electron microscope (FE-TEM) (Tecnai G2 F20, FEI, Hillsboro, OR, USA), field emission scanning electron microscope (FE-SEM) (SU8020, Hitachi, Tokyo, Japan) images were obtained for structural and morphological characterization of TiO 2 films, PC 61 BM layers and perovskite films. The current-voltage (J-V) curves were obtained under standard simulated air-mass (AM) 1.5 sunlight generating from a solar simulator (Sol 3A, Oriel, Newport, RI, USA). UV-vis absorption spectra were characterized by ultroviolet visible absorption spectrometer (Avantes, Apeldoom, The Newtherlands). Photoluminescence (PL) spectrum was measured with PL testing system (LabRAW HR800, HORIBA Jobin Yvon, Paris, France). All the characterization of devices was performed in ambient atmosphere at room temperature.

Results and Discussion
In this paper, we investigate the effect of low temperature process of conventional TiO 2 -based ETL on the performance of PSCs. Hence, we have characterized the structure of TiO 2 nanoparticles with TEM, as demonstrated in Figure 1, showing the average size of about 5 nm. Firstly, we investigate the influence of thermodynamic process on the growth and formation of nano TiO2 films. The detailed information about experiments is shown in the second paragraph of experimental Section 2.2.2. The SEM images are shown in Figure 2. The J-V curves of PSCs and the corresponding data extracted from J-V curves are shown in Figure 3 and Table 1, respectively. Firstly, we investigate the influence of thermodynamic process on the growth and formation of nano TiO 2 films. The detailed information about experiments is shown in the second paragraph of experimental Section 2.2.2. The SEM images are shown in Figure 2. The J-V curves of PSCs and the corresponding data extracted from J-V curves are shown in Figure 3 and Table 1, respectively.
It is clearly observed from Figure 2a that there are large holes remained in nano TiO 2 film prepared with Direct method. Figure 2b shows that these holes locate at valleys of FTO where the surface curvature changes greatly. Some voids can still be discovered from Figure 2c and d obtained from nano TiO 2 film prepared under Proportional method, however, they are smaller compared with Figure 2a,b. Figure 2f shows the FTO is effectively covered by nano TiO 2 film with few defects fabricated under Delayed method. We can also observe that the nano TiO 2 layer and FTO surface fit well with each other, indicating the conformal growth of nano TiO 2 on FTO substrate. Firstly, we investigate the influence of thermodynamic process on the growth and formation of nano TiO2 films. The detailed information about experiments is shown in the second paragraph of experimental Section 2.2.2. The SEM images are shown in Figure 2. The J-V curves of PSCs and the corresponding data extracted from J-V curves are shown in Figure 3 and Table 1, respectively. It is clearly observed from Figure 2a that there are large holes remained in nano TiO2 film prepared with Direct method. Figure 2b shows that these holes locate at valleys of FTO where the surface curvature changes greatly. Some voids can still be discovered from Figure 2c and d obtained from nano TiO2 film prepared under Proportional method, however, they are smaller compared with Figure 2a,b. Figure 2f shows the FTO is effectively covered by nano TiO2 film with few defects fabricated under Delayed method. We can also observe that the nano TiO2 layer and FTO surface fit well with each other, indicating the conformal growth of nano TiO2 on FTO substrate.
We can conclude that the morphology of nano TiO2 film deposited on FTO substrate can be controlled by optimizing annealing process and nano TiO2 film exhibits high quality when Delayed method was introduced, which is ascribed to the slower evaporating rate of the residual solvent, resulting in conformal covering of TiO2 over FTO substrate. Generally, films with fewer cracks and defects endow the PSCs with better performance, which is also confirmed by reverse scanning J-V curves of the devices in Figure 3 and data extracted from reverse scanning J-V curves in Table 1. We can conclude that the morphology of nano TiO 2 film deposited on FTO substrate can be controlled by optimizing annealing process and nano TiO 2 film exhibits high quality when Delayed method was introduced, which is ascribed to the slower evaporating rate of the residual solvent, resulting in conformal covering of TiO 2 over FTO substrate. Generally, films with fewer cracks and defects endow the PSCs with better performance, which is also confirmed by reverse scanning J-V curves of the devices in Figure 3 and data extracted from reverse scanning J-V curves in Table 1.    Secondly, as shown in Figure 2, the TiO 2 nanoparticles are not connected tightly with each other, which may lead to leakage of PSCs. Therefore, we studied the effect of PC 61 BM on the interface between nano TiO 2 layer and perovskite layer. The detailed information about experiments is shown in the third paragraph of experimental Section 2.2.2. The SEM images are shown in Figure 4. covered by PC61BM and we can hardly find the light islands, leaving few cracks. While the concentration further rises to 25 mg/mL, the nano TiO2 layer was fully covered by PC61BM, giving smooth and uniform film.
As demonstrated in previous report [23], it is too hard for thick ETL to transfer the electron from the absorption layer to FTO substrate and instead relatively thin ETL leads to a serious recombination of electrons and holes. Hence, we can infer that the device optimized with PC61BM of 15 mg/mL exhibits the best performance, in which PC61BM filled the most valleys of nano TiO2 layer, leaving few cracks, preventing the contact between FTO substrate and perovskite layer and improving the electron transporting efficiency.   Finally, to investigate the influence of PC61BM substrate on the formation and properties of perovskite films, we studied the SEM images and absorbance spectra of perovskite films deposited on PC61BM of different concentration (5, 10, 15, 20, 25 mg/mL), absorbance spectra of PC61BM of different concentration and PL spectra of perovskite films deposited on PC61BM of different concentration, as shown in Figures 5, 6a,b and 7, respectively.
As shown in Figure 5, when the PC61BM precursor solution concentration is lower than 15 mg/mL, a good grained and uniform perovskite films were obtained. In contrast, obviously different perovskite surface morphologies are observed on 20 and 25 mg/mL PC61BM substrate, where the crystallinity of the perovskite films is distinctly deteriorated with serious grain boundary distortion and large holes.
From Figure 6a, we can observe that the absorption spectra signify the similar light-harvesting capabilities over 400 to 900 nm regardless of the different PC61BM concentration adopted. All five films exhibit approximately the same absorption edge at about 790 nm with no red-shift or blue-shift, corresponding to optical bandgap of perovskite. Particularly, perovskite films deposited on 25 mg/mL PC61BM manifest inferior light-harvesting capabilities, which is caused by the internal defects resulted from crystal distortion, and it is consistent with Figure 5j. From Figure 6b, we can observe that the PC61BM exhibits higher light-harvesting capabilities with the concentration increasing. However, perovskite films deposited on 15, 20 and 25 mg/mL PC61BM manifest similar absorbance, which indicates that light-harvesting capabilities of perovskite films deposited on PC61BM layers decrease with the precursor concentration (≥15 mg/mL) of PC61BM. As shown in Figure 4d,f,h,j,l, it is clearly observed that there are some light islands distributed on the dark substrate. We can determine that the light islands are peaks of nano TiO 2 layer and the dark substrate is PC 61 BM comparing with Figure 4b. From the magnified SEM images, we can conclude that the islands disappear gradually with the concentration of PC 61 BM increasing from 5 to 25 mg/mL. When the concentration of PC 61 BM is 5 mg/mL, the PC 61 BM just filled the valleys of nano TiO 2 layers. With the concentration of PC 61 BM increased to 15 mg/mL, the nano TiO 2 film was well covered by PC 61 BM and we can hardly find the light islands, leaving few cracks. While the concentration further rises to 25 mg/mL, the nano TiO 2 layer was fully covered by PC 61 BM, giving smooth and uniform film.
As demonstrated in previous report [23], it is too hard for thick ETL to transfer the electron from the absorption layer to FTO substrate and instead relatively thin ETL leads to a serious recombination of electrons and holes. Hence, we can infer that the device optimized with PC 61 BM of 15 mg/mL exhibits the best performance, in which PC 61 BM filled the most valleys of nano TiO 2 layer, leaving few cracks, preventing the contact between FTO substrate and perovskite layer and improving the electron transporting efficiency.
Finally, to investigate the influence of PC 61 BM substrate on the formation and properties of perovskite films, we studied the SEM images and absorbance spectra of perovskite films deposited on PC 61 BM of different concentration (5, 10, 15, 20, 25 mg/mL), absorbance spectra of PC 61 BM of different concentration and PL spectra of perovskite films deposited on PC 61 BM of different concentration, as shown in Figures 5-7, respectively. shift, corresponding to optical bandgap of perovskite. Particularly, perovskite films deposited on 25 mg/mL PC61BM manifest inferior light-harvesting capabilities, which is caused by the internal defects resulted from crystal distortion, and it is consistent with Figure 5j. From Figure 6b, we can observe that the PC61BM exhibits higher light-harvesting capabilities with the concentration increasing. However, perovskite films deposited on 15, 20 and 25 mg/mL PC61BM manifest similar absorbance, which indicates that light-harvesting capabilities of perovskite films deposited on PC61BM layers decrease with the precursor concentration (≥15 mg/mL) of PC61BM.  quenching of excitons resulted from the following two reasons, radiative relaxation of excited electrons back to the ground state of perovskite and electron injection from light-absorption layer into electron transport layer [24]. Hence, we can infer that the PC61BM of 15 mg/mL deposited on nano TiO2 layer guaranteed the high-quality interface between ETL and perovskite layer, resulting in more efficient injection of electrons from perovskite layer into nano TiO2 layer, leading to low PL intensity, which is consistent with Figure 5f. We also have performed J-V characteristics of PSCs optimized with PC61BM, as shown in Figure  8 and Table 2. From Figure 8 and Table 2, we can discover that following the increasing of the concentration, the PCE extracted from reverse scanning J-V curves shows up a different trend about approximately 0.81%, 2.52%, 5.48%, 3.14% and 0.26% from the perovskite solar cells with PC61BM concentration of 5, 10, 15, 20 and 25 mg/mL respectively. It is clearly observed that the PCE reaches the maximum value when the concentration is increased up to 15 mg/mL, and then decreases with the further increasing of the concentration, which is consistent with the above discussion. As shown in Figure 5, when the PC 61 BM precursor solution concentration is lower than 15 mg/mL, a good grained and uniform perovskite films were obtained. In contrast, obviously different perovskite surface morphologies are observed on 20 and 25 mg/mL PC 61 BM substrate, where the crystallinity of the perovskite films is distinctly deteriorated with serious grain boundary distortion and large holes.
From Figure 6a, we can observe that the absorption spectra signify the similar light-harvesting capabilities over 400 to 900 nm regardless of the different PC 61 BM concentration adopted. All five films exhibit approximately the same absorption edge at about 790 nm with no red-shift or blue-shift, corresponding to optical bandgap of perovskite. Particularly, perovskite films deposited on 25 mg/mL PC 61 BM manifest inferior light-harvesting capabilities, which is caused by the internal defects resulted from crystal distortion, and it is consistent with Figure 5j. From Figure 6b, we can observe that the PC 61 BM exhibits higher light-harvesting capabilities with the concentration increasing. However, perovskite films deposited on 15, 20 and 25 mg/mL PC 61 BM manifest similar absorbance, which indicates that light-harvesting capabilities of perovskite films deposited on PC 61 BM layers decrease with the precursor concentration (≥15 mg/mL) of PC 61 BM. Figure 7 shows PL spectra of perovskite films deposited on PC 61 BM of different concentration. It is clearly observed that the peak of photoluminescence is fairly weak when nano TiO 2 layer has not been optimized with PC 61 BM, being explained by the existence of via holes in nano TiO 2 layer, which leads to leakage. Furthermore, from Figure 7, we can observe that the PL intensity decrease as the concentration of PC 61 BM increasing from 5 to 15 mg/mL. However, when the concentration of PC 61 BM is further increased to 25 mg/mL, the PL intensity raises up again and reaches the maximum value.
As demonstrated in a previous report, the PL intensity of the films is closely related to quenching of excitons resulted from the following two reasons, radiative relaxation of excited electrons back to the ground state of perovskite and electron injection from light-absorption layer into electron transport layer [24]. Hence, we can infer that the PC 61 BM of 15 mg/mL deposited on nano TiO 2 layer guaranteed the high-quality interface between ETL and perovskite layer, resulting in more efficient injection of electrons from perovskite layer into nano TiO 2 layer, leading to low PL intensity, which is consistent with Figure 5f.
We also have performed J-V characteristics of PSCs optimized with PC 61 BM, as shown in Figure 8 and Table 2. From Figure 8 and Table 2, we can discover that following the increasing of the concentration, the PCE extracted from reverse scanning J-V curves shows up a different trend about approximately 0.81%, 2.52%, 5.48%, 3.14% and 0.26% from the perovskite solar cells with PC 61 BM concentration of 5, 10, 15, 20 and 25 mg/mL respectively. It is clearly observed that the PCE reaches the maximum value when the concentration is increased up to 15 mg/mL, and then decreases with the further increasing of the concentration, which is consistent with the above discussion.     Finally, we demonstrate the carbon electrode-based flexible solar cell of PCE of 3.24% and 2.60% for reverse and forward scanning, respectively [25]. The carbon electrode is fabricated with spongy film composed of carbon nanoparticles, which leads to unstable contact in characterization, resulting in poor device performance. The golden electrode-based device is under investigation. Moreover, we are investigating the optimization process for perovskite layer and hole transport layer and the optimization process for the interface between layers to improve the device performance.

Conclusions
We have discovered a method to improve the performance of PSCs by optimizing low-temperature annealing process of solution-processed TiO 2 ETL. In this work, we have received better nano TiO 2 films using Process 1c (the as-grown nano TiO 2 film had been delayed for one hour before the annealing temperature was raised from room temperature to 150 • C at rate of 8 • C/min). Moreover, the surface of nano TiO 2 film optimized with 15 mg/mL PC 61 BM endows the device with better performance. Finally, we have demonstrated the carbon electrode-based flexible perovskite solar cell, and the golden electrode-based device is under investigation. Our results can serve as a platform to fabricate high-efficiency flexible PSCs and have a potential in flexible and wearable devices.
Author Contributions: X.Y. wrote the paper. D.C. designed the experiments. X.Y., X.Z. and J.C. analyzed the data. Y.Y. and C.C. prepared the samples. B.L., J.W., G.L. and Z.Z. performed all the measurements. X.Z. supervised the project. All authors commented and approved the paper. All authors have read and agreed to the published version of the manuscript.