High efficiency perovskite solar cells using DC sputtered compact TiO2 electron transport layer

High conductivity and transparency of the electron-transporting layer (ETL) is essential to achieve high efficiency perovskite solar cells (PvSCs). Generally, titanium dioxide (TiO2) has been extensively utilized as an ETL in PvSCs. Both surface roughness and uniformity of the compact-TiO2 (C-TiO2) can influence the efficiency of the PvSC. This work investigates the optimization of the direct current (DC) sputtering power and the ratio of argon (Ar) to oxygen (O2) plasma to achieve high quality ETL films. The effect of changing the DC sputtering power on the C-TiO2 films and subsequently on the overall efficiency was studied. The electrical and optical properties of the C-TiO2 layer were characterized for various DC powers and different ratios of Ar to O2 plasma. It was found that the optimum preparation conditions for the C-TiO2 films were obtained when the DC power was set at 200W and a flow rate of 6 sccm Ar and 12 sccm O2. A power conversion efficiency (PCE) of 15.3% in forward sweep and 16.7% in reverse sweep were achieved under sunlight simulator of 100mW/cm. These results indicate that significant improvement in the efficiency can be achieved, by optimizing the C-TiO2 layer.


Introduction
Due to the continued global energy demands, alternative energy sources that are sustainable and less harmful to the environment are being explored. The organic-inorganic hybrid perovskite solar cells (PvSCs) have been first developed in 2009 with conversion efficiency of 4% [1]. One decade later, it has reached an efficiency of over 25% [2]. The efficiency of PvSCs has been continuously improving by employing different materials for both the active layer and electron and hole transport layers as well as different fabrication techniques. Researches have been focused on how to improve the composition of the perovskite film to yield high quality crystal structure [3], and how to improve the stability and PCE [4,5]. Although there are noticeable improvements in the PvSCs performance, its stability and reproducibility are still impeding market penetration and their wide scale commercialization. Generally, both the ETL and the hole transport HTL layers forms an interfacing layers in PvSCs. The organic inorganic perovskite layer is sandwiched between the cathode (ETL), and anode (HTL) layers [6,7]. Thus, those interface layers are essential in any effort to improve the performance of the PvSCs. One of the critical layers that can determine the photovoltaic performance of the PvSCs is the ETL. The ETL plays a central function in transferring electrons to the external electrodes and blocking the backflow of the holes at the same time. Therefore, it is worth optimise the ETL and understand its effects on cell stability and efficiency. Although there are a number of materials such as SnO 2 , ZnO that can be utilized as an ETL, most of the ETL materials are based on the titanium oxide (TiO 2 ) due to its good conductivity and high transparency. For example, a typical value of the SnO 2 transparency is about 90% and its conductivity is around 600 S/cm at room temperature [8]. Whereas, the transparency of the TiO 2 is 90% over the visible range with much higher conductivity of 2.2 Â 10 4 S/cm [9].
In addition to C-TiO 2 , mesoporous TiO 2 (M-TiO 2 ) is one of most widely used ETL for the PvSCs, it has good stability, high transparency, excellent optoelectronic properties, and short deposition processes [10]. In addition, the M-TiO 2 layer can block holes and collect electrons because of its large surface area and ability to extract electrons by improving the interface layer [11]. Moreover, an effective balance between the flux of holes and electrons can be achieved using the M-TiO 2 . The effective balance between holes and electronics flow in the mesoporous structured PvSCs has resulted in less current-voltage hysteresis compared with other types of the ETL. However, the transmission path of electrons in the C-TiO 2 layer is much shorter than in the M-TiO 2 layer [12]. Improved performance of the PvSCs can be achieved by optimising the ETL to prevent critical carrier recombination at the interfaces.
Liu et al. has proposed using Li-doped C-TiO 2 layer and achieved PCE of 17% [11]. A number of studies reported on the optimisation and improvements of the C-TiO 2 layer to enhance the efficiency of the PvSCs [10,13,14]. However, most of these methods are complicated and require high temperature treatment. For example, Saliba et al. in 2018 [15] demonstrated a 20% efficiency of PvSCs by using C-TiO 2 prepared using aerosol spray pyrolysis and lithium salt doping of the M-TiO 2 . This technique requires high temperature treatment by holding the sample at 450°C for 1 h. Another attempt in the same year, Sidhik et al. has studied the effect of using cobalt as doping for the M-TiO 2 , resulting in efficiency of 18% [16]. The ETL in their study consisted mainly of two layers. The first layer was a blocking TiO 2 and the second layer was M-TiO 2 treated with different concentrations of FK209. This method required two high temperature treatments, the first by holding the sample at 500°C for 1 hour and the second annealing the sample at 450°C for 30 min.
Recently in 2021, Jeong et al. has reported on perovskite solar cells with power conversion efficiency of 25.6%, these cells have long-term operational stability (450 h) and show intense electroluminescence with external quantum efficiencies of more than 10%. Their ETL was prepared using high temperature process; "Prior to the spraying process, the FTO substrates were placed on a hot plate and the temperature was increased to 450°C rapidly. After the spray pyrolysis step, the substrates were stored at 450°C for 1 h and then slowly cooled to room temperature. The FTO/C-TiO 2 substrates prepared with m-TiO 2 were heated at 500°C on a hot plate for 1 h to remove organic compounds first, and then slowly cooled to 200°C" [2].
In the present work, DC-sputtering was used to deposit the C-TiO 2 . The DC-sputtering is a well establish technique employed for high throughput scalable manufacturing environment. It is a low temperature processing, which does not require holding the sample at elevated temperatures as compared with the aerosol spray pyrolysis of C-TiO 2 deposition method. Moreover, the perovskite film reported by Sidhik  There is no study, to our knowledge, that reported on the use of C-TiO 2 deposited by DC-sputtering with the M-TiO 2 layer as an ETL. A combined C-TiO 2 and cobalt FK209 doped M-TiO 2 were used as an ETL. Both cobalt FK209 doped M-TiO 2 were spun coated after the deposition of C-TiO 2 . The effect of different DC sputtering powers and ratios of argon (Ar) to oxygen (O 2 ) plasma during the deposition of the C-TiO 2 were investigated.
Finally, the influence of preparation conditions on the J sc , V oc , FF, and PCE of PvSCs were observed. Furthermore, the reproducibility and stability of the cell were improved compared with other deposition methods.
2 Experimental details

Solar cell fabrication 2.2.1 Substrate preparation
In this work, a commercially available FTO (FTO, 12-15 V/sq) glass substrate was utilized. Practically, the glass substrate coated with FTO was etched one-third of its total area to allow for front contact formation. The zinc powder (Sigma-Aldrich) and dilute hydrochloric acid (2 M) diluted in deionized (DI) water were mixed as an etching for the FTO coated substrate to avoid short circuits during the characterization of the cell. Substrates were cleaned in the ultrasonic bath using acetone, methanol, and isopropanol for 10 min, 15 min, 5 min respectively. Finally, the substrates were dried with pure nitrogen and kept in the oven for 1 h at 100°C. Oxygen plasma Asher was used for 10 min to clean and remove any residual organic particles on the substrates. This is to improve the adhesion between the FTO and subsequent layers.

Preparation of electron transport layers (ETL)
Two main layers were used in this work as an ETL, a C-TiO 2 and M-TiO 2 . Firstly, the C-TiO 2 layer was deposited using DC-sputtering with a titanium target (3" diameter Â 0.25" thick and 99.995% purity). The DC-sputtering source is powered by 200 W and the process plasma were a mixture of Argon (Ar) and Oxygen (O 2 ) with flow rates of 6 and 12 standard cubic centimetres per minute (sccm) respectively. The C-TiO 2 film was deposited onto the FTO substrate with thicknesses between 60 and 70 nm forming an amorphous TiO 2 film. The second layer is called the mesoporous TiO 2 (M-TiO 2 ), which was prepared by using 150 mg of TiO 2 paste (30N-RD) dissolved in 1 mL of ethanol. This liquid mixture is stirred vigorously overnight at temperature of 70°C. Doping of M-TiO 2 was obtained by adding cobalt Fk209 salt to further enhance the electronic transport layer properties. The doping solution was prepared by mixing cobalt (2.5 mg) with acetonitrile (200 mL) then added to the M-TiO 2 solution. Cobalt doped M-TiO 2 was deposited by spin coating resulted in 150 nm thick films based on setting the spinner at a speed of 4000 rpm with 2000 rpm/s acceleration for 10 s. After finishing the spin coating process, the film was dried for a few minutes by preheating at 100°C on hot plate. The temperature was increased gradually to 125°C then held the samples for 5 min. The heating cycle was repeated by increasing the temperature to 375°C for 15 min and held for 5 min. Finally, increasing the temperature to 450°C then keep samples on the hot plate for 30 min to form an anatase phase during the annealing process. All the annealing processes were performed at the laboratory ambient. Practically, a glass lid was used to cover the samples on the hot plate during the whole annealing process to enable a more temperature uniformity distribution over the substrate. Once the above process is finished, the samples were left on the hot plate to gradually cool down to 125°C before moving them to the glove box to deposit the perovskite layer at room temperature. The samples were kept in a pure nitrogen circulated glove box.

Perovskite solar cell fabrication
The temperature inside the glove box was kept between 21 and 27°C as higher temperatures above 28°C might affect the quality of perovskite layer. Temperatures higher than 28°C during film preparation is known to influence the crystallization of perovskite [17], and thus might affect the perovskite film quality [15]. The active perovskite layer was prepared by a method described elsewhere [18]. Table 1 shows the main materials used to prepare the perovskite precursor.
Principally, anhydrous dimethylformamide (DMF) and dimethyl sulfoxide (DMSO) were prepared by the volume ratio of DMF: DMSO 4:1 (v:v).The chemical formula of the perovskite material used in this study is CsI 0.05 [(FAPbI 3 ) 0.85 (MAPbBr 3 ) 0.15 ] 0.95 . Using the concentration in Table 1, the solution precursor was prepared. The CsI is prepared by dissolving 1.5 M of CsI in DMSO. Hence, a 5% ratio of the C S I was added to the total precursor to attain the required "triple-cation composition" [18]. The final stock of (5% C S I and solution of perovskite precursor) inside the glovebox was deposited by two steps spin coating: first step, was dropping 100 mL of final stock on the sample (TiO 2 covered sample), then using a low spinning speed of 1000 rpm for 10 s to ensure full surface coverage. A second step uses a high speed 6000 rpm spinning for 25 s was followed. After 15 s, a 220 mL of chlorobenzene (CBZ) was quickly poured by pipet in the middle of the 2.5 Â 2.5 cm 2 sample ten seconds before the end of the spinning. This step is critical as longer or shorter times can results in poor quality films. As a result, both of (DMSO/DMF) have been removed from the substrate by CBZ during spin coating. After finishing the preparation process, perovskite film seems semi-transparent and was annealed at 100°C for 40-50 min.

Hole transport layer (HTL) and backside electrode
The HTL was prepared by dissolving 80 mg of spiro-MeOTAD (Sigma-Aldrich) in 1mL of CBZ. Then adding 29 mL of 4-tert-butylpyridine (TBP) to the spiro-MeOTAD mixture. A 520 mg of lithium salt bis (trifluoromethane) sulfonimide was dissolved in 1 mL of acetonitrile, and a 300 mg of cobalt salt (FK209) salt was dissolved in 1 mL of acetonitrile. Finally, adding 18 mL of Li salt TFSI and 30 mL of cobalt salt (FK209) to the stock solution of spiro-MeOTAD as additives. A 100 mL of the final stock solution is deposited at the beginning of the spinning rotation at 4000 rpm for 20 s. This layer should look homogeneously distributed and smooth over the sample area due to using dynamic deposition method. Then, 70 nm of gold was deposited by electron beam evaporator as the electrodes. In this work, a shadow mask (heat resistance paper) was used to form a pattern of Au electrode onto the cell. Figure 1 illustrates the schematic for PvSCs fabrication procedure followed in this work.  The cross-section of the sample was obtained using scanning electron microscope (SEM) based on a Raith-150 EBL machine. Atomic force microscope (AFM) digital instruments (Veeco Instruments Inc. DI 3100) and the Raith SEM were used to characterize the morphology of the C-TiO 2 . Cary 6000i ultraviolet-visible spectrometer was utilized to measure the C-TiO 2 layer absorbance and transparency versus wavelength.

Device and measurement characterization
3 Results and process discussion

Structure properties of compact TiO 2
To understand the influence of the C-TiO 2 on the performance of the PvSCs, the C-TiO 2 film was deposited using DC-sputtering at different powers. Four film layers prepared with different DC powers (180, 200, 220 and 240 watt) were deposited on FTO substrate and studied. The electrical and electronic properties of these layers were measured using van-der-Pauw four-probe method. It can be seen from Table 2 that the conductivity of the C-TiO 2 sputtered at 200 W is higher than other layers sputtered at different DC powers. Besides, the mobility of the C-TiO 2 layer sputtered at 200 W is also higher compared with the other preparation conditions. The conductivity of the C-TiO 2 sputtered at 200 W was measured for samples prepared at different ratios of Ar and O 2 . It can be noticed from Table 3 that the optimum ratio of Ar and O 2 is equal to 6 and 12 sccm respectively. This optimum ratio of Ar/O 2 achieved high conductivity value of 1.6 Â 10 3 [S/cm] and lower resistivity of 6.15 Â 10 À4 [V.cm].
Since the properties of the C-TiO 2 layer play a significant role in the overall performance of the PvSCs, the C-TiO 2 surface topography were investigated with AFM and SEM imaging. Figure 3 shows the surface of the C-TiO 2 sputtered at 200 W has denser grains as compared with samples sputtered at (180, 220, 240) watts. Furthermore, the C-TiO 2 film sputtered at 200 W resulted in a smoother surface and was more homogeneous compared with other layers. This is demonstrated by the low surface roughness of the "root mean square" (RMS). The surface roughness value of the C-TiO 2 sputtered at 200 W is 17 nm, which is lower than films deposited at different DC powers.   It is found that the C-TiO 2 sputtered at 200 W gave improved performance as it provided smother interface with the M-TiO 2 layer. The AFM and SEM images in Figure 3 show that grains size of the DC sputtered C-TiO 2 film powered at 200 W is smaller than the grains size of the other layers. Based on the C-TiO 2 powered at 200 W, grain sizes between 100-400 nm were obtained with an average size of 250 nm.

Current-voltage characterization
All current voltage characteristics were conducted using (ABET Sun3000) sunlight simulator at AM 1.5G and an average light intensity of 100 mw/cm 2 . Samples were 2.5 Â 2.5 cm 2 in size with an average active area of 0.36 cm 2 and measurements were taken at room temperature of 21-27°C. The active area is defined by the area of the window opening on the otherwise masked backside of the glass substrate. Measured value of the active device area is 0.36 cm 2 which is larger than most of the reported areas in the literatures [12][13][14][15][16]. The cells were exposed to simulated sunlight through this window. Furthermore, we are using FTO glass substrate with sheet resistance of 12-15 V/sq. When all of these factors were taken into considerations, the average forward and reverse scan of J sc were 24.5 mA/cm 2 . This is consistent with the highest J sc obtained with PvSC [2].
All the results discussed in this study were conducted on active area of 0.36 cm 2 , however, for comparison, cells with active area of 0.5 cm 2 and 1 cm 2 were also fabricated. The forward and reverse scan efficiency of the 0.5 cm 2 and 1 cm 2 were 15%, 16.5% and 11.5%, 12.1% respectively. The efficiency of the 1 cm 2 device is considerably lower than the 0.36 cm 2 device due to increased defect density, however, scaling up is beyond the scope of this work. An Abet technology model 15150 reference cell with KG5 filter traceable to NIST, NREL, Fraunhofer ISE and ISPRA standard artefacts were employed to certify our efficiency measurements. The full characterisation of the reference cell is given in the supplementary section. Figure 5 displays the (J-V) characteristic curves of the PvSCs based on C-TiO 2 DC sputtered at (180, 200, 220, 240) watt. All fabrication parameters of the PvSC including the DC powers used for the deposition of the C-TiO 2 are shown in Table 4. It is demonstrated that the PvSCs based on sputtered C-TiO 2 and cobalt doped M-TiO 2 as an ETL achieved the highest efficiency with an average PCE of 16%. The measurement was repeated a few times and averaged with resulting current density (J sc ) of 24.5 mA/cm 2 , fill factor (FF) of 61.4% and open-circuit voltage (V oc ) of 1021 mV. The C-TiO 2 layer deposited using DC sputtering at 200 W gave the highest efficiency of 15.3% in the forward and 16.7% in the revers scan compared to solar cells prepared at other DC powers. The PCE has improved by 17% in forward scanning and 11% in backward scanning for the C-TiO 2 sputtered at 200 W compared to the C-TiO 2 sputtered at 240 W. There is an increase of about 19% in forward scan and 6% in reverse scan in PCE for C-TiO 2 sputtered at 200 W compared with the C-TiO 2 sputtered at 180 W. This is in agreement with the resistivity decrease in layers deposited at different powers as shown in Table 2. The variations in short circuit current density were small between samples and, averaged around 24 mA/cm 2 in forward and backward scanning but it has dropped to about 22 mA/cm 2 for C-TiO 2 sputtered at 180 W. The V oc is maintained at about 1 V for all types in the forward scanning, and was more than 1 V in the backward scanning. The FF has improved by 3.4% and 8.3% in forward scanning and 2.3% and 11% in backward scanning for the C-TiO 2 prepared at 200 W compared to the C-TiO 2 sputtered at 180 W and 240 W respectively. The combination of the DC sputtered compact TiO 2 ETL with the Cesium Lead Iodide CsI 0.05 [(FAPbI 3 ) 0.85 (MAPbBr 3 ) 0.15 ] 0.95 based perovskite offered the potential for scaling up and resulted in a more reproducible cells with good stability [16]. In addition, we attribute this improvement in PCE to the drop in the resistivity of the C-TiO 2 sputtered at 200 W layer comparing to others as shown in Table 2. Figure 6a shows transparency of FTO glass covered with C-TiO 2 having a thickness of 70 nm. It is worth noting that the C-TiO 2 has approximately the same transparency as the bare FTO glass. It is clear that the transparency of the C-TiO 2 prepared with different DC powers has more than 90% transparency over the visible wavelength range. Figure 6b presents the absorbance of the C-TiO 2 measured using Ultraviolet-Visible spectrometry (Cary 6000i). From the curves shown in Figure 6 there is a weak absorption between 350 and 800 nm, and much higher absorption between 250 and 350 nm in accordance with the expected absorption spectra for TiO 2 .   Figure 7 shows that the highest transparency was achieved when of the ratio of O 2 to Ar was 6 to 12 sccm.

Stability and reproducibility of the perovskite solar cell
To examine the reproducibility of our process, we have prepared a total of 20 solar cells all with DC sputtered C-TiO 2 layers. Ten samples were fabricated using DC sputtering powered at 200 W and 5 samples at each of the other DC powers. The stability of the fabricated devices were studied by measuring the solar cell parameters (EFF%, FF%, V oc , J sc , V max , J max , I sc , R shunt , and R series ) under sunlight simulator condition (AM 1.5G [100 mW/cm 2 ] using [ABET Sun3000] and was monitored over 70 days period. Figures 8  and 9 illustrate the forward and reverse scan efficiency measurements conducted over the length of 10 weeks in two different atmospheres, in a vacuum held dissector (Fig. 8) and in laboratory ambient (Fig. 9).
The drop in efficiency was from 2% to 5% during the 70 days period measured at one week intervals. Ten samples were tested for stability studies and some samples were kept in a desiccator and compared with samples kept in laboratory ambient for stability measurements. The average drop for all samples was 3.4%. Not all of the samples efficiency dropped at the same rate, so the figure below shows the variation in efficiency as measured weekly.
As prepared efficiency of champion device in the forward and reverse scanning were 15.3% and 16.7% respectively. Four weeks later, the efficiency in the forward and reverse scanning dropped to 15% and 16.5%   respectively. After eight weeks, the efficiency in the forward and reverse scanning was about 14.6% and 16% respectively. The percentage of drop in efficiency is maintained over the 70 days period the test was conducted. We did not observe any large variations in the efficiency or variations from sample to sample when prepared under same conditions an indication that the use of DC sputtered compact TiO 2 offer greater reproducibility and stability compared to other techniques. However, samples stored in ambient condition exhibit more percentage drop in efficiency of 7% as compared to those kept in a dissector were an average of 3.4% drop is measured in agreement with the trend observed in perovskite based solar cells.

Conclusion
We have investigated the role of DC power and the Ar to O 2 ratio in sputtered TiO 2 to form compact TiO 2 layer suitable for high efficiency perovskite solar cells. An optimum sputtering condition to obtain the desired TiO 2 layer was found to be 200W DC power with Ar (6 sccm) and O 2 (12 sccm). The cobalt doped M-TiO 2 were deposited on C-TiO 2 to further improve the electron transport layer (ETL) through increasing its conductivity. The combined DC sputtered C-TiO 2 and cobalt doped M-TiO 2 has resulted in solar cells efficiencies higher than 16%. It has been found that the C-TiO 2 sputtered at 200 W possess high conductivity and carrier mobility compared to C-TiO 2 sputtered at other DC powers. Moreover, the surface roughness of the C-TiO 2 sputtered at 200 W is almost 38% smaller than the C-TiO 2 prepared at other DC powers. Perovskite material deposited on the optimised C-TiO 2 and cobalt doped M-TiO 2 produced solar cells with higher, J sc , V oc , FF and efficiencies as high as 15.3% in forward and 16.7% at backward scanning. Cells prepared at these conditions have shown good reproducibility and an average drop of 5% in efficiency over 70 days exposure to standard sunlight at room temperature.

Author contribution statement
A. Hayali conceived the original idea, performed the experimental work and wrote draft of the manuscript. M. M. Alkaisi supervised the work, performed analysis and contributed to the discussion of the results, and revised the manuscript.