The Investigation for Coating Method of Titanium Dioxide Layer in Perovskite Solar Cells

: The e ﬀ ect of conventional Perovskite solar cells (PSCs) by using di ﬀ erent concentration and spin-coating speeds of titanium dioxide (TiO 2 ) as an electron transport layer (ETL) was studied. The inﬂuence of TiO 2 based on device structure: ﬂuorine-doped tin oxide substrate / TiO 2 / Perovskite (CH 3 NH 3 PbI 3 ) / 2,2 (cid:48) ,7,7 (cid:48) -Tetrakis[N,N-di(4-methoxyp phenyl)amino]-9,9 (cid:48) -spirobiﬂuorene / silver, is also studied. The spin-coating speed is varied in a range from 1000 to 3000 rpm to get optimal performance of device. The optimized power conversion e ﬃ ciency (PCE) of PSCs with original concentration (OC) and double concentration (DC) TiO 2 is 8.74 and 9.93%, respectively. The reason is attributed to excellent absorption in shorter wavelength, compact characteristic, and suitable thickness of TiO 2 , leading to perfect short-circuit current density (Jsc), lower series resistance (Rs), and higher ﬁll factor (FF) of 0.75. Besides, recombination of electron and hole is also decreased due to the compact feature, leading to higher open-circuit voltage (V OC ) of 0.91 V.


Introduction
People are dedicated to develop and advance photovoltaics because solar source is an emerging renewable energy with progress of technology and energy depletion. The outstanding breakthrough for solar cells is the discovery of perovskite in the Ural Mountains of Russia by Gustav Rose in 1839 and is named after Russian mineralogist Lev Perovski. Perovskite solar cells (PSCs) are a type of solar cells containing a perovskite structured compound and considered as light-harvesting active layer, most commonly a hybrid organic-inorganic lead or tin halide-based material [1][2][3][4][5][6].
In the past few years, lead halide perovskites have gained tremendous attention around the globe because of their excellent photovoltaic characteristics, including high charge carrier mobility [1], large absorption coefficient [2], and tunable optical properties [3]. Organic-inorganic hybrid solar cells based 2 of 9 on lead halide perovskites have achieved a phenomenal improvement in power conversion efficiency (PCE) from 3.8% to 24.02% in ten years. The remarkable record of high PCE for perovskite solar cells can be recorded 3.8% published by Miyasaka et al. took organometal halide perovskites as sensitizers in liquid electrolyte-based dye-sensitized solar cells (DSSCs) in 2009 [4]. In 2011, Park et al. realized an efficiency of 6.5% by applying TiO 2 surface treatment prior to deposition [7]. In 2012, an all-solid-state lead iodide perovskite sensitized TiO 2 mesoscopic heterojunction solar cell with long-term stability had achieved a PCE of 9.7% [8]. In 2013, an outstanding PCE of 15% received from perovskite-sensitized mesoscopic solar cells was prepared using the sequential deposition method [9]. The astonished breakthrough of outstanding PCE is to reach 20.1%, published by Bakr et al. in 2015 [5]. Moreover, it is worth mentioning that there are some research groups devoted to apply perovskite solar cells on flexible substrates to develop wearable devices for illumination, exercise, etc., [10,11]. The DSSCs has been studied by a lot of researchers since Gratzel reported a DSSC device with sandwich-mode architecture in 1991 [12]. The traditional DSSC structure includes a dye-sensitized photoanode, a counter electrode, and an electrolyte [13]. The photoanode includes a TiO 2 porous film and a TiO 2 compact film [14]. The TiO 2 compact layer is conventionally prepared by many methods as following: aerosol spray pyrolysis [15][16][17], thermal oxidation [18], spin-coating method [14], atomic layer deposition [19], and electrochemical deposition [20]. Besides, TiO 2 film is a promise candidate to act as an electron transport layer (ETL) in PSCs according to best absorption at short wavelengths [21][22][23].
In this study, we discuss the performance of PSCs with different concentration and various spin-coating speeds of TiO 2 to understand the fundamental effect. The TiO 2 solution is prepared with original and double concentration, however, grown TiO 2 film is also fabricated at 1000, 2000, and 3000 rpm. The properties of TiO 2 film and improved PSCs are also researched to further optimize PCE.

Materials and Methods
In this paper, all materials were purchased by commercial source and used without further sublimation to accomplish fabrication. All preparation of solution and thin film was continuously achieved in glove box. The cleaning process of substrate was referred and complied with standard operation procedure of previous researches. The work function of fluorine-doped tin oxide-coated glass substrates (FTO-glass, with a sheet resistance of 7 Ω/sq, AimCore Technolog, Hsinchu, Taiwan) was improved by treatment of oxygen (O 2 ) plasma with 10 W for 2 min [24] to reduce carbon contamination and effectively grow subsequent thin film.

Fabrication of Thin Film and Device
The TiO 2 precursor solution was dynamically spin-coated onto FTO substrates at different coating speeds (1000/2000/3000 rpm) for 60 s and heated at 180 • C for 5 min on hotplate in air to grow TiO 2 thin Crystals 2020, 10, 236 3 of 9 film. TiO 2 thin film was annealed at 550 • C for 30 min to form compact characteristic. The perovskite layer prepared by mixing solvent was deposited using solvent-engineering technique [25]. Perovskite precursor solution was spin-coated onto TiO 2 /FTO substrate in two consecutive spin-coating steps at 1000 rpm of 10 s and 5000 rpm of 20 s in order. During second step, toluene was dropping onto the wet spinning film. Perovskite thin film was annealed at 90 • C for 15 min on hotplate before hole transport layer was deposited by spin-coating spiro-OMeTAD solution at 3000 rpm for 30 s. Finally, silver (Ag) metal as anode, which is defined by shadow mask with active area of 6 mm 2 , was deposited to thicknesses of 100 nm by vacuum thermal evaporation below pressure of 4.8 × 10 −6 torr. The deposition rate was monitored and adjusted by quartz crystal oscillator to ensure film thickness

Characteristic Measurement
All measurements were carried out in air. The current density-voltage (J-V) characteristics of device were measured with power source meter (Keithley 2400, Keithley, Cleveland, AL, USA) under an illumination of 100 mW/cm 2 with an AM 1.5G sun simulator (Oriel 96000 150W Xe lamp, Newport, Taipei, Taiwan). The light intensity was calibrated by reference solar cell and meter (Oriel 91150, Newport, Taipei, Taiwan). The spectrum was taken for −5 V sample bias to separate sample and secondary edge for analyzer. The samples were measured by field-emission scanning electron microscopy (SEM, JEOL JSM-7001, Japan) and X-ray diffraction (XRD, ultima IV, Rigaku Corporation, Tokyo, Japan) to demonstrate cross-section images and crystalline structure, respectively. Devices were further measured by atomic force microscope (AFM, XE-70, Park Systems, Suwon, Korea) to analyze surface morphology with scan rate of 0.5 Hz probing an area of 10 × 10 µm 2 , with effectively reducing speed at which the tip moves and allowing system more time to adjust to sudden changes in height, and by spectrophotometer (UV-3900, Hitachi, Tokyo, Japan) to analyze the absorption spectrum and transmittance of devices. The external quantum efficiency (EQE, QE-3000, Titan Electro-Optics, Taipei, Taiwan) measurement was performed on irradiation of 150 W xenon light tower in DC mode.

Results and Discussion
In this study, both energy band diagram and structure are shown in Figure 1. Conventional structure of PSCs is fabricated as following: FTO/TiO 2 /perovskite (CH 3 NH 3 PbI 3 )/spiro-OMeTAD/Ag in Figure 1a. The transportation of electron can be clear to understand owing to suitable lowest unoccupied molecular orbital of TiO 2 in Figure 1b. Figure 2 shows current density-voltage (J-V) characteristic of devices obtained by TiO 2 with original concentration (OC) and double concentration (DC) at various spin-coating speeds (1000, 2000, and 3000 rpm), corresponding to red, blue, and purple line, respectively. We can clearly see red line have a colossal area under intersection of xand y-coordinate compared to others. Generally speaking, the fill factor (FF) is determined from the intersection of area when J-V curve has been achieved. It is obviously shown that device performance prepared by OC TiO 2 is not improved when coating speed is increased by two and three times in Figure 2a. The same situation and results occur in device prepared by DC TiO 2 in Figure 2b. The optimized power conversion efficiency (PCE) of device prepared by OC and DC TiO 2 is 8.74 and 9.93% with 1000 rpm, respectively. Poor PCE of device prepared by OC and DC TiO 2 is only 0.77 and 1.2% with 3000 rpm, respectively. The higher series resistance (Rs) (6290 Ωcm 2 and 728 Ωcm 2 ) and lower FF (0.14 and 0.15) are one of the active factors to cause poor performance in Table 1. The Table 1 presents detailed performance of all devices. It can convey information about differences between OC and DC TiO 2 , and effect on various speed of spin-coating fabrication. As coating speed decreases one times, the PCE of device prepared by OC TiO 2 is enhanced from 4.38 to 8.74% because of a great increase in short-circuit current density (J SC ) from 7.58 to 19.21 mA/cm 2 , indicating that the ability of electron transport is enhanced when thickness of TiO 2 gradually increases. Similar improvement of PCE is obtained by DC TiO 2 , from 7.4 to 9.93%. However, it is seen that both JSC and Rs are affected in Table 1, indicating a two times increase in TiO2 concentration can cause improvement in PCE. Then, JSC of device is improved from 7.58 to 16.29 mA/cm 2 at 2000 rpm and from 6.51 to 9.31 mA/cm 2 at 3000 rpm, respectively. It is also shown that all Rs results caused by TiO2 concentration of two times give an individual drop at each speed from 282 to 107 Ωcm 2 , from 173 to 142 Ωcm 2 , and from 6290 to 728 Ωcm 2 , respectively. In order to find the reason why TiO2 influences the performance of PSCs, we decided to measure EQE spectrum and compare the two devices with highest PCE in Figure 3.  However, it is seen that both JSC and Rs are affected in Table 1, indicating a two times increase in TiO2 concentration can cause improvement in PCE. Then, JSC of device is improved from 7.58 to 16.29 mA/cm 2 at 2000 rpm and from 6.51 to 9.31 mA/cm 2 at 3000 rpm, respectively. It is also shown that all Rs results caused by TiO2 concentration of two times give an individual drop at each speed from 282 to 107 Ωcm 2 , from 173 to 142 Ωcm 2 , and from 6290 to 728 Ωcm 2 , respectively. In order to find the reason why TiO2 influences the performance of PSCs, we decided to measure EQE spectrum and compare the two devices with highest PCE in Figure 3.   However, it is seen that both J SC and Rs are affected in Table 1, indicating a two times increase in TiO 2 concentration can cause improvement in PCE. Then, J SC of device is improved from 7.58 to 16.29 mA/cm 2 at 2000 rpm and from 6.51 to 9.31 mA/cm 2 at 3000 rpm, respectively. It is also shown that all Rs results caused by TiO 2 concentration of two times give an individual drop at each speed from 282 to 107 Ωcm 2 , from 173 to 142 Ωcm 2 , and from 6290 to 728 Ωcm 2 , respectively. In order to find Crystals 2020, 10, 236 5 of 9 the reason why TiO 2 influences the performance of PSCs, we decided to measure EQE spectrum and compare the two devices with highest PCE in Figure 3. of wavelength has greatest effect. The black and red curve represent the optimized device of OC and DC TiO2, respectively, reaching a maximum of 74.27% in 576 nm and 79.27% in 234 nm. Two curves have increased significantly from 300 to 390 nm and decreased until 800 nm. From 390 to 610 m, black curve can reach about 71.33% of EQE in average. From 380 to 650 nm, red curve can reach about 76.66% of EQE in average. The difference between current density from EQE data is close to that obtained from J-V curves in Table 1. Therefore, the current density data are reliable to explain why PCE of DC TiO2 device is higher than that of the OC one. Furthermore, we use absorption and transmission spectrum to further describe the relationship between light importance and device performances.   EQE spectrum of optimized device is demonstrated in Figure 3. Figure 3 displays which range of wavelength has greatest effect. The black and red curve represent the optimized device of OC and DC TiO 2 , respectively, reaching a maximum of 74.27% in 576 nm and 79.27% in 234 nm. Two curves have increased significantly from 300 to 390 nm and decreased until 800 nm. From 390 to 610 m, black curve can reach about 71.33% of EQE in average. From 380 to 650 nm, red curve can reach about 76.66% of EQE in average. The difference between current density from EQE data is close to that obtained from J-V curves in Table 1. Therefore, the current density data are reliable to explain why PCE of DC TiO 2 device is higher than that of the OC one. Furthermore, we use absorption and transmission spectrum to further describe the relationship between light importance and device performances. Figure 4 illustrates the UV-vis absorption and transmission spectrum of device prepared by OC and DC TiO 2 ETL at various spin-coating speeds in visible range from 300 nm to 800 nm, respectively. All curves can be divided into two parts, in which black (1000 rpm), red (2000 rpm), and blue lines (3000 rpm) represent PSCs obtained by OC TiO 2 at different speeds, and green (1000 rpm), pink (2000 rpm), and brown lines (3000 rpm) represent that of DC one. It is obvious that absorption intensity of DC TiO 2 parts are higher than that of OC TiO 2 , leading to better capability to absorb light source and then convert it to PCE. As we know, TiO 2 is a wide band gap semiconductor, which has a strong absorption at short wavelengths [23,24]. The order of intensity for all curves with peak position are arranged as following: black (0.724 a.u. in 300 nm), green (0.662 a.u. in 312 nm), red (0.549 a.u. in 300 nm), pink (0.543 a.u. in 312 nm), brown (0.473 a.u. in 312.5 nm), and blue (0.291 a.u. in 300.5 nm). Compared to the average intensity of absorption spectrum, DC TiO 2 film is 0.559 a.u., which is higher than 0.521 a.u. of OC TiO 2 film. Therefore, more light has been absorbed when PSCs are fabricated by DC TiO 2 ETL. Further cross-section images of optimized PSCs obtained by DC TiO 2 ETL are indicated by measurement of SEM in Figure 5. The thickness of each layer can be demonstrated and cross-section image of SEM is further measured via rotate and zero angle. The surface roughness (Rq) and morphologies of OC and DC TiO 2 film coated on FTO substrate are shown in Figure 6. The difference of Rq is too close (maximum and minimum value is 29.892 and 22.254 nm, respectively) and all values are listed in Table 2. As shown in Figure 6a-d, these surfaces are compact due to island distribution, compared to the result of columnar distribution in Figure 6e,f. Figure 7a shows the XRD spectra of OC TiO 2 film fabricated on FTO substrate. No extra peak of OC TiO 2 film is produced due to basic peaks of FTO substrate, indicating the ability of electron transportation is affected by surface roughness at interface between Perovskite and TiO 2 . However, although the difference of surface roughness is not huge in Table 2, morphology of DC TiO 2 film at Crystals 2020, 10, 236 6 of 9 1000 rpm is more compact to others in Figure 6d. The unexpected crystallite growths because a small peak of DC TiO 2 film is found at 2 θ = 26 • , which represents tetragonal anatase structure (JCPSD card no. 83-2243) in Figure 7b.
We ascribe the growth factor of tetragonal anatase structure to annealing temperature of 90 • C but there is no sufficient evidence to demonstrate [25]. So, the reason for PCE improvement is attributed not only to absorb light contribution of short wavelength from TiO 2 film but effective blocking of recombination between electron and hole in TiO 2 film, leading to enhancement of open-circuit (V OC ) from 0.85 to 0.91 V. When spin-coating speed decreases, the thickness is gradually thinner and the ability of electron transportation is also increased, corresponding to boost J SC and FF in Table 1. Therefore, better PCE of 8.74% and 9.93% are achieved with thinner OC and DC TiO 2 at 1000 rpm (average value of 55.33 nm), respectively. In the future, we predict that the thickness and crystalline characteristic of compact layer is the most important factor to affect the performance of PSCs because charge transport would be largely enhanced by suitable TiO 2 layer. obtained from J-V curves in Table 1. Therefore, the current density data are reliable to explain why PCE of DC TiO2 device is higher than that of the OC one. Furthermore, we use absorption and transmission spectrum to further describe the relationship between light importance and device performances.    Figure 4 illustrates the UV-vis absorption and transmission spectrum of device prepared by OC and DC TiO2 ETL at various spin-coating speeds in visible range from 300 nm to 800 nm, respectively. All curves can be divided into two parts, in which black (1000 rpm), red (2000 rpm), and blue lines (3000 rpm) represent PSCs obtained by OC TiO2 at different speeds, and green (1000 rpm), pink (2000 rpm), and brown lines (3000 rpm) represent that of DC one. It is obvious that absorption intensity of DC TiO2 parts are higher than that of OC TiO2, leading to better capability to absorb light source and then convert it to PCE. As we know, TiO2 is a wide band gap semiconductor, which has a strong absorption at short wavelengths [23,24]. The order of intensity for all curves with peak position are arranged as following: black (0.724 a.u. in 300 nm), green (0.662 a.u. in 312 nm), red (0.549 a.u. in 300   Figure 7a shows the XRD spectra of OC TiO2 film fabricated on FTO substrate. No extra peak of OC TiO2 film is produced due to basic peaks of FTO substrate, indicating the ability of electron transportation is affected by surface roughness at interface between Perovskite and TiO2. However, although the difference of surface roughness is not huge in Table 2, morphology of DC TiO2 film at 1000 rpm is more compact to others in Figure 6d. The unexpected crystallite growths because a small peak of DC TiO2 film is found at 2 θ = 26°, which represents tetragonal anatase structure (JCPSD card no. 83-2243) in Figure 7b.      Figure 7a shows the XRD spectra of OC TiO2 film fabricated on FTO substrate. No extra peak of OC TiO2 film is produced due to basic peaks of FTO substrate, indicating the ability of electron transportation is affected by surface roughness at interface between Perovskite and TiO2. However, although the difference of surface roughness is not huge in Table 2, morphology of DC TiO2 film at 1000 rpm is more compact to others in Figure 6d. The unexpected crystallite growths because a small peak of DC TiO2 film is found at 2 θ = 26°, which represents tetragonal anatase structure (JCPSD card no. 83-2243) in Figure 7b.

Conclusions
The PSCs have been fabricated with different concentration (OC and DC TiO 2 ) and with various spin-coating speeds. The effect of TiO 2 on electron transportation has been studied via photovoltaic characteristic with AFM and XRD in this study. It is found that the main reason for highest PCE of PSCs obtained by OC and DC TiO 2 is attributed to light absorption at a shorter wavelength and more compact surface of TiO 2 film, indicating the improvement of J SC and FF, respectively. The contribution can be further shown in absorption spectra and surface roughness. Although there is no enough information to prove why and how extra crystalline of 26 • in XRD is grown, it is still another factor for increase in PCE. The recombination of electron and hole also decreases according to compact DC TiO 2 film, leading to enhancement of V OC 0.85 to 0.91 V. Therefore, optimized TiO 2 film as ETL in PSCs have two features as following: thin and compact. Finally, the slight change in concentration of TiO 2 have attained excellent performance of PSCs when we find the suitable parameters and details of improved thin films.