Organic-inorganic hybrid perovskite solar cells (PSCs) are thirdgeneration solar cells that are proving to be promising replacements for silicon solar cells. Despite their simple manufacturing process at a low temperature, PSCs have recently achieved their best efficiency of 25.7 %, which is comparable to that of silicon solar cells [1]. Titanium dioxide (TiO₂) films have been widely used as electron transport layers (ETLs) for sensitized solar cells because of their thermal and chemical stability, as well as a band alignment suitable for solar cells [2]. Several methods are available for TiO₂ thin film fabrication, such as solgel, sputtering, chemical vapor deposition, and spray coating [3–7]. While the sol-gel method is inexpensive and can produce a uniform thin film [8], it is difficult to control the thickness and apply it to a large-area application because the sol-gel method is based on the spincoating method [9]. Furthermore, the heat treatment process involved can lead to crystallization [9] as well as the formation of cracks or pinholes due to thermal stress [10,11]. Therefore, a magnetron sputtering method has been proposed as a solution [12–15]. In this method, TiO₂ is formed through sputtering by reacting a TiO₂ ceramic target or a Ti pure metal target with oxygen [16]. Generally, ceramic targets are expensive and fragile, and direct current (DC) power cannot not be used due to the movement restriction of electrons in non-metallic materials [17]. However, for reactive sputtering (RS) using the Ti metal target, both DC power and radiofrequency (RF) power can be utilized. The RF sputtering method facilitates the adjustment of the film thickness through a slow deposition process [14], and the energy of the sputtered ions reaching the substrate helps the formation of a uniform thin film [18,19]. Moreover, sputtering is performed at a lower temperature than that in other methods, so that it can be applied when using a substrate that is sensitive to heat [20].

In this work, we propose the application of RFRS TiO₂ to multiple cation/anion CsFAMAPb(IBr)₃ (CFM)-based PSCs. Although RS with a metal target generally uses DC power, we used RF power in this experiment. This is because a TiO₂ phase is formed on the surface of the Ti target during DC sputtering, making it difficult to deposit a uniform and reproducible TiO₂ thin film. The RFRS TiO₂ layer was optically and electrically analyzed, and then compared with general spin-coated TiO₂. Structural characteristics were evaluated by scanning electron microscopy (SEM) and X-ray diffraction (XRD) analysis. The optical transmittance of the TiO₂ layers was measured by UV-Vis spectroscopy. Since RFRS produces a thin and uniform TiO₂ layer, it led to better device performance depending on the post-annealing treatment. In addition, RFRS TiO₂ successfully blocked indium diffusion from the indium tin oxide (ITO) layer during high-temperature annealing. Finally, an improved power conversion efficiency (PCE) of 17.46 % was achieved by using RFRS TiO₂ in the CFM-based PSCs.

2.1. Sample preparation

The TiO₂ films were prepared on 1.5 × 1.5 cm² ITO-coated glass substrates. The ITO substrates were cleaned prior to coating by using an ultrasonicator with surfactants, deionized water, acetone, and ethanol sequentially, followed by blow-drying with N₂ gas. The RFRS TiO₂ was deposited on the substrate using a Ti metal target (99.99 %, Materion) with an RF power of 80 W under 2.5 sccm O₂ gas flow. The base and working pressures were 5.0 × 10−8 Torr and 3.0 mTorr, respectively. For the spin-coated TiO₂, the cleaned ITO substrate was further UV/ozone treated for 30 min. The TiO₂ solution was then spin-coated onto the treated ITO substrate at 7,000 rpm for 8 s, 1,000 rpm for 10 s, and 2,000 rpm for 40 s, and then dried at 125 °C for 5 min [21]. The spin-coating procedure was repeated twice to ensure the appropriate thickness, following which the post-annealing treatment was performed at 450 °C for 1 h in the furnace. The TiO₂ solution consisted of 1:0.15 mol of 1-butanol (99.8 %, Sigma Aldrich) and titanium diisopropoxide bis(acetylacetonate) (TTIP, 75 wt% in isopropanol, Sigma Aldrich), and used after stirring for 2 h.

2.2. Device fabrication

The structure of the CFM based PSCs was Au/Spiro-OMeTAD/CFMbased perovskite/TiO₂/ITO on glass. A perovskite precursor solution was prepared, which contained the following solvents: lead(II) iodide (PbI₂, 99.999 %, Sigma Aldrich), lead(II) bromide (PbBr₂, > 98.0 %, TCI), formamidinium iodide (FAI, Greatcellsolar), methylammonium bromide (MABr, Greatcellsolar), and cesium iodide (CsI, > 99.0 %, TCI) in N,N-dimethylformamide (DMF, 99.8%, Sigma Aldrich) and dimethyl sulfoxide (DMSO, 99.9 %, Sigma Aldrich). This mixture was stirred for 2 h, following which it was coated on the TiO₂/ITO substrate using an anti-solvent method [21]. Chlorobenzene (CB, > 98.0 %, TCI) was used as the anti-solvent. For the hole-transporting layer (HTL), a spiro-OMeTAD solution was spin-coated on the substrate at 3,000 rpm for 30 s. The spiro-OMeTAD solution was prepared by mixing spiro-OMeTAD powder with chlorobenzene (CB, 99.8 %, Sigma Aldrich), 4-tert-butylpyridine (89 %, Sigma Aldrich), bis(trifluoromethane) sulfonimide lithium salt (99.95 %, Sigma Aldrich), and acetonitrile (99.8 %, Sigma Aldrich). Finally, the 100-nm Au electrode was deposited with an active area of 0.1 cm² by thermal evaporation.

2.3. Device characterization

The transmittance of the TiO₂ films was measured using a UV-Vis spectrometer (Hitachi, UH4150). The surface morphology and thickness of the TiO₂ layers were monitored by field-emission scanning electron microscopy (FE-SEM, Hitachi, SU8220). The TiO₂ structure was confirmed by using a high-resolution X-ray diffractometer (HR-XRD, Malvern Panalytical, Empyrean). The indium concentration and depth profile of the TiO₂ layer were analyzed using an Xray photoelectron spectrometer (Ulvac-PHI, Japan). Sheet resistance measurements were carried out using a 4-point probe system (AIT, SR1000N), and the performance of the solar cell was measured using a solar simulator under air mass (AM) 1.5 G solar illumination (class AAA) equipped with a 450 W Xenon Arc Lamp 6280NS.

Figure 1(a) shows an RF reactive magnetron sputtering system using a Ti metal target. Radio waves ionized the inert gas (Ar) through plasma, following which the ionized Ar atoms bombarded the target material. The sputtered atoms were deposited on the ITO substrate, where they reacted with active O₂ gas to form a TiO₂ layer coating the substrate. Figure 1(b) shows the transmittance spectra of spincoated TiO₂ after annealing for 1 h at 450 °C (black solid line) and as-deposited RFRS TiO₂ (red solid line). RFRS TiO₂ and spin-coated TiO₂ showed a similar transmittance (75–80 %) in the range between 350 to 900 nm. As seen in the SEM images in Fig. 2, the thickness of the spin-coated TiO₂ was approximately 50 nm [Fig. 2(b)], while the thickness of the RFRS TiO₂ was 20 nm, which was less than half of that of the spin-coated TiO₂ film [Fig. 2(d)]. Despite this difference in thickness, the similar transmittance values of the two samples implied that the post-annealed spin-coated TiO₂ contains more anatase structures than does the RFRS TiO₂ [22]. However, as the thickness of TiO₂ increased, the series internal resistance increased and short-circuit current density decreased [23]; thus, we expect that RFRS deposition of a thin TiO₂ layer would positively impact device performance. The transmittance curve of RFRS TiO₂ annealed under the same condition as spin-coated TiO₂ (450 °C for 1 h) is shown in Fig. 1(b). As shown in Fig. 1(b), the transmittance of RFRS TiO₂ (red solid line) seems to be slightly improved compared to before annealing due to the change in crystallization from amorphous to anatase structure. Additionally, the surface of the spin-coated TiO₂ is more granular and has a multitude of pinholes on the surface as shown in Fig. 2(a). In contrast, the RFRS TiO₂ film has a denser and more uniform surface [Fig. 2(c)], which improves the hole blocking capability and shunt resistance [24,25]. Figures 2(e) and 2(f) show the top-view and cross-sectional SEM images of RFRS TiO₂ layer annealed at 450 °C for 1h, respectively. Even after annealing at 450 °C for 1h, the morphology was uniform without significant damage and the thickness was maintained at 20 nm.

Figure 1. (a) Schematic illustration of TiO₂ deposition process with RF reactive magnetron sputtering system. (b) Transmittance spectra of spin-coated TiO₂ (black solid line), RFRS TiO₂ (red solid line), and RFRS TiO₂ after annealing for 1 h at 450 °C (red dashed line).

Figure 2. SEM images of (a) top-view spin-coated TiO₂, (b) cross-sectional CFM based perovskite/spin-coated TiO₂, (c) top-view RFRS TiO₂, (d) cross-sectional CFM based perovskite/RFRS TiO₂, (e) top-view RFRS TiO₂ after annealing for 1 h at 450 °C, and (f) cross-sectional CFM based perovskite/RFRS TiO₂ after annealing for 1 h at 450 °C (Scale bar: 200 nm).

Before characterizing the device performance, we performed XRD analysis to investigate the structure of TiO₂ corresponding to each method. All the TiO₂ samples were prepared on an ITO substrate. The spin-coated TiO₂ showed an XRD peak at 25.5° [Fig. 3(a)]. Diffraction pattern at 25.5° is related to the anatase (101) TiO₂ structure [26]. Therefore, the main peak at 25.5° signifies that the anatase structure is dominant in the spin-coated TiO₂, which is consistent with the transmittance results in Fig. 1. In contrast, as-deposited RFRS TiO₂ showed no signal initially, but after annealing at 450 °C for 1 h, a new peak appeared at 25.5° [Fig. 3(b)]. Since the anatase TiO₂ structure is generally formed at high temperatures, it can be concluded that a transition from the amorphous to the anatase phase occurred during the postannealing process in RFRS TiO₂ [27].

Figure 3. (a) XRD patterns of spin-coated TiO₂ after 450 °C annealing for 1 h (black, bottom), as-deposited RFRS TiO₂ (blue, middle), and RFRS TiO₂ after annealing for 1 h at 450 °C (red, top) (∗: anatase (101) TiO₂ stucture). (b) XRD patterns magnified in the 20−30∘ region for RFRS TiO₂ with and without annealing process.

Further, the annealing treatment may cause indium diffusion at the TiO₂-ITO interface as well as structural conversion of TiO₂ [28]. ITO has been widely used as a solar cell substrate owing to its high transmittance and low roughness [29], but it is affected by temperature to a greater extent than is fluorine-doped tin oxide (FTO) [30]. Figure 4(a) illustrates the XPS depth profile, which represents the indium atomic concentration according to the depth of the TiO₂/ITO structure. Since there is a difference in thickness between the spin-coated TiO₂ layer and the RFRS TiO₂ layer, the monitoring depth was normalized for the respective thickness of each film. In the case of spin-coated TiO₂, the indium atomic concentration increased in the TiO₂ region (green dots) post-annealing at 450 °C. In contrast, the RFRS TiO₂ displayed a near-zero indium concentration near the ITO region despite the heat treatment at 450 °C; this result indicates that the RFRS TiO₂ was sufficiently dense to block indium diffusion. We also measured the sheet resistance of the ITO substrate depending on the annealing temperature, as shown in Fig. 4(b). After annealing at 250, 350, and 450 °C, the sheet resistance of ITO tended to increase to 35 Ωsq⁻¹ as the temperature increased, and the color of ITO became slightly yellow, as shown in the inset of Fig. 4(b). In particular, the resistance change increased steeply when the temperature was increased from 250 to 350 °C. The increase in sheet resistance is known to be mainly due to a reduction in oxygen vacancies [31]. In other words, the increase in sheet resistance with increasing temperature was caused by both indium diffusion as well as the reduction of oxygen vacancies in the TiO₂/ITO structure.

Figure 4. (a) XPS depth profiles indicating indium concentration in spin-coated TiO₂ after annealing process (black square), as-deposited RFRS TiO₂ (red circle), and RFRS TiO₂ after annealing process (blue triangle). (b) Sheet resistance of ITO substrates depending on the annealing temperature.

Finally, we characterized device performance using the spin-coated TiO₂ and RFRS TiO₂; the device structure was Au/Spiro-OMeTAD/ CFM based perovskite/TiO₂/ITO, as illustrated in the upper inset of Fig. 5(a). In addition, the statistical efficiency data of PSCs are shown in the lower inset of Fig. 5(a). The current density-voltage (J-V) curves were measured in the reverse scan direction under standard AM 1.5 G solar illumination (100 mW cm^-2). The device parameters of CFMbased PSCs fabricated with spin-coated TiO₂ and RFRS TiO₂ ETLs are summarized in Table I. The open-circuit voltage (V_OC), short-circuit density (J_SC), fill factor (FF), and PCE for the device with spin-coated TiO₂ were 1.017 V, 21.625 mA/cm², 72.902 %, and 16.044 % respectively. The device parameters with RFRS TiO₂ were 1.028 V, 21.555 mA/cm², 74.269 %, and 16.463 % after annealing at 450 °C for 1 h. The solar cell with spin-coated TiO₂ showed slightly higher J_SC but lower V_OC and FF values than the device with RFRS TiO₂. The difference in J_SC can be explained by the spin-coated TiO₂ having a higher amount of anatase structures than RFRS TiO₂. The V_OC and FF values are greatly influenced by the interlayer interfaces related to shunt resistance [24,25,32], and RFRS TiO₂ has a more uniform surface compared to spin-coated TiO₂, as mentioned SEM data in Fig. 2. Moreover, spin-coated TiO₂ may have a larger series resistance due to its greater thickness; hence, the device with RFRS TiO₂ shows higher V_OC and FF values. In addition, since indium diffusion affects the FF, the spin-coated TiO₂ had a reduced FF due to the larger number of pinholes.

Table 1 . Device parameters for perovskite solar cells with spin-coated and RFRS TiO₂ ETL after annealing for 1 h at 450 °C..

Devices	PCE (%)	VOC (V)	JSC (mA/cm-2)	FF (%)
Spin-coated TiO2	16.044	1.017	21.625	72.902
RFRS TiO2	16.463	1.028	21.555	74.269

Figure 5. Comparison of J-V curves between spin-coated TiO₂ (black square) and RFRS TiO₂ (red circle) after annealing at 450 °C. The upper and lower insets are the schematic diagram of the device architecture and the statistical efficiency graph of the PSCs, respectively.

Figure 6 shows the J-V curves of the CFM-based PSCs with RFRS TiO₂ based on their annealing temperatures. The inset in Fig. 6 shows the statistical efficiency of the PSCs fabricated with RFRS TiO₂ layer at different annealing tempeatures. The device parameters are summarized in Table II. The device with as-deposited RFRS TiO₂ showed V_OC and FF values of 1.068 V and 76.353 %, respectively. These values increased to 1.095 V and 78.605 % after heat treatment at 250 °C. However, when the heat treatment was performed at temperatures above 250 °C, these values tended to decrease again; as the temperature increased further, the PCE gradually decreased from 17.457 to 16.422 %. The best device performance was achieved at an annealing temperature of 250 °C for 1 h.

Table 2 . Device parameters for perovskite solar cells with RFRS TiO₂ depending on the annealing temperatures..

Devices	PCE (%)	VOC (V)	JSC (mA/cm-2)	FF (%)
Before annealing	16.547	1.068	20.301	76.353
250 °C	17.457	1.095	20.276	78.605
350 °C	16.469	1.045	20.791	75.811
450 °C	16.422	1.054	21.074	73.955

Figure 6. J-V curves of devices using RFRS TiO₂ layers at different annealing temperatures (inset: the staticalcal efficiency graph of the PSCs for each annealing temperature).

We applied an RFRS TiO₂ ETL to CFM-based PSCs and systematically characterized it in comparison with spin-coated TiO₂. The RFRS TiO₂ had a thickness of 20 nm, with a transmittance comparable to 75−80 % in the visible range. After annealing, the amorphous phase was converted to an anatase structure in RFRS TiO₂. In the case of spin-coated TiO₂, indium diffusion and an increase in sheet resistance with the annealing temperature were observed. However, the RFRS TiO₂ had a denser and more uniform surface, because of which it could block indium diffusion, thereby increasing the V_OC and FF values. Finally, the CFM-based PSCs with RFRS TiO₂ improved the device performance of the perovskite solar cell, with a PCE of 17.46 %. We believe that our results demonstrate the potential of RFRS TiO₂ for use as an ETL as well as suggest effective directions for future PSC applications.

This research was supported by the Korea Electric Power Corporation (Grant number: R20XO02-8).