Scalable Pulsed Laser Deposition of Transparent Rear Electrode for Perovskite Solar Cells

Sputtered transparent conducting oxides (TCOs) are widely accepted transparent electrodes for several types of high‐efficiency solar cells. However, the different sputtering yield of atoms makes stoichiometric transfer of target material challenging for multi‐compounds. Additionally, the high kinetic energies of the arriving species may damage sensitive functional layers beneath. Conversely, pulsed laser deposition (PLD) is operated at higher deposition pressures promoting thermalization of particles. This leads to stoichiometric transfer and additionally reduces the kinetic energy of ablated species. Despite these advantages, PLD is rarely used within the photovoltaic community due to concerns about low deposition rates and the scalability of the technique. In this study, wafer‐scale (4‐inch) PLD of high‐mobility Zr‐doped In2O3 (IZrO) TCO for solar cells is demonstrated. IZrO films are grown at room temperature with deposition rate on par with RF‐sputtering (>4 nm min−1). As‐deposited IZrO films are mostly amorphous and exhibit excellent optoelectronic properties after solid phase crystallization at <200 °C. 100‐nm thick films feature a sheet resistance of 21 Ω◻−1 with electron mobilities ≈70 cm2 V−1s−1. PLD‐grown IZrO is applied as rear electrode in efficient semi‐transparent halide perovskite solar cells leading to the improved stabilized maximum power point efficiency (15.1%) as compared to the cells with sputtered ITO electrodes (11.9%).

pulses of ns duration. This leads to explosive removal of the material from the target surface confined in the plasma plume that expands in the direction perpendicular to the substrate. The ablation process typically takes place in the background gas environment which serves the purpose of thermalizing the energetic species, as well as providing additional supply of elements (e.g., oxygen for the deposition of oxides) to compensate for lack of stoichiometry. The plasma plume is significantly smaller than the 4-inch substrate leading to non-uniform thickness distribution when collected statically on substrates bigger than 1-inch. Therefore, the laser spot is scanned over the target to obtain homogeneous films on larger substrates. In addition to the plume spatially scanning over the 4-inch wafer in the lateral direction, the substrate holder is rotated to ensure a complete coverage of the plume over the full wafer.
The precise physics underlying the deposition process is more complex than sketched here and is thoroughly reviewed in references. [14,15] Nevertheless, it is important to highlight the main advantages of PLD.
1. Allowing stoichiometric transfer of elements independent of their volatility and evaporation pressures. 2. Providing flexibility for processing parameters since the energy source for material ablation is physically decoupled from the vacuum equipment. This gives a larger choice for deposition pressures as no restriction for glow discharge pressure is present. This also allows control of particles' kinetic energy when landing on the substrate. 3. Allowing the precise control of the number of arriving particles enabling layer-by-layer growth thanks to the pulsed nature of the process.
Atomic layer engineering, mentioned in point 3, is greatly associated with in-situ growth monitoring with reflection highenergy electron diffraction (RHEED) [16] allowing monolayer growth control. However, we do not further elaborate on this feature here, as it is less relevant for the growth of polycrystalline films discussed in this work. On the other hand, we discuss other above-mentioned advantages by comparing the properties of the PLD-grown and sputtered films of IZrO from the same target composition.

Zr-Doped In 2 O 3 Fabrication and Analysis
The IZrO films were grown on glass substrates at room temperature using a large area PLD system (Twente Solid State Technologies) with capabilities for deposition on up to 4-inch wafers. Glass substrates were ultrasonicated in acetone and isopropanol for 5 minutes and rinsed in deionized water prior to a deposition. A KrF excimer laser (λ = 248 nm) was used for all experiments with the repetition rate of 20 Hz and the fluence of 1.9 J cm −2 .
The total working pressure was 0.02 mbar with the optimized oxygen to total flow ratio r(O 2 ) = O 2 /(Ar+O 2 ) of 20%. It is important to mention that this working pressure is almost one order of magnitude higher than typically used in magnetron sputtering (0.0013 mbar as equivalent of 1 mTorr, for instance). This is expected to slow down the arriving species and enable "soft landing" on top of sensitive transport layers [5,17,18] and as briefly discussed in Section 4.4.
Deposition rates for the complex oxide growth is commonly estimated in nm per pulse as certain amount of pulses corresponds to one monolayer of the deposited material and can be monitored by intensity oscillation of the RHEED pattern. This is not directly informative for the growth of polycrystalline IZrO as the same deposition rate per pulse would correspond to different thicknesses if the repetition rate is varied. Therefore, it is estimated that the deposition rate as the overall thickness divided by the deposition time. 2.5 × 10 4 pulses leads to ≈100 nm of the deposited IZrO film. This, assuming a commonly used repetition rate of 10 Hz, corresponds to the deposition rate of ≈2.5 nm min −1 . The deposition rate was further improved to ≈5 nm min −1 increasing the repetition rate to 20 Hz without sacrificing the optoelectonic properties.
The uniformity of a typical PLD IZrO film (≈120 nm) across the horizontal axis on a 4-inch substrate is displayed in Figure 1b. The thickness of the surface step height on samples placed across the x-axis was measured by a stylus profilometer with the tolerance of ±3% of 120-nm film thickness due to the tip available. Normalized thickness was calculated as www.advmattechnol.de the ratio between the thickness at the measured point and its mean value across the x-axis of the wafer. Overall, thickness uniformity of ±5.3% on the full wafer is obtained. Maximum variation of ≈10% is observed in the center area (which can also be a single measurement outlier) and would be further addressed by optimizing the plume-scan curve. Moreover, solar cell precursors with 9 cm 2 area were placed closer to the 4-inch substrate edges as demonstrated in Section 4.4. The local thickness uniformity in this 30 mm region is ±2.5%. Thicknesses of films were verified by X-ray reflectivity measurements as displayed in Figure S2, Supporting Information. Clearly visible interference fringes indicate highly smooth surfaces of IZrO layers given the large density contrast between glass and In 2 O 3 . Fitting of the measured specular reflectivity scans demonstrate root mean squared surface roughness below 0.75 nm for all IZrO samples (PLD-grown and sputtered TCO films both in as-deposited and annealed state as would be elaborated below).
To study the evolution of the structural and optoelectronic properties of the IZrO films with temperature, as well as for full comparison with sputtered IZrO films, as-deposited IZrO electrodes were post-annealed at 200 °C in air for 30 min. This temperature budget is not compatible for films deposited on top of the perovskite absorber layer. However, the enhanced optoelectronic performance after the heat treatment step (as shown in Section 4.1) is beneficial if IZrO film is applied as front electrode in four-terminal (4T) tandem devices or in silicon heterojunction cells with a thermal budget of 200 °C. [19] Comparison of properties was done with IZrO films grown by RF-sputtering at KAUST following procedure described by Aydin et al. [20] All films analyzed were deposited (both by PLD and sputtering) from a 98/2 wt% In 2 O 3 /ZrO 2 target. Microstructure of the films was inspected by grazing incidence X-ray diffraction (GI-XRD) with the incident angle (ω) = 0.6°. A heating stage was used for the in situ study of amorphous-to-polycrystalline transition. The electrical resistivity (ρ), carrier concentration (N e ), and Hall mobility (μ Hall ) were extracted from Hall effect measurements in the Van der Pauw configuration. The optical transmittance (T) and reflectance (R) of the films were measured by UV-vis-NIR spectrometer with an integrating sphere. Absorptance (A) was determined from 100−T−R.

Solar Cell Fabrication and Analysis
Semi-transparent perovskite solar cells (ST-PSC) were prepared inside the glovebox with N 2 environment and the oxygen and moisture levels at about 1 ppm. 9 cm 2 area Corning XG glass substrates were cleaned sequentially with soap, deionized water and isopropanol in an ultrasonic bath prior to the front electrode fabrication. Sn-doped In 2 O 3 (ITO) was deposited at room temperature using an AJA RF magnetron sputtering system from a 90/10 wt% In 2 O 3 /SnO 2 target onto glass substrates for the front contact and on top of the perovskite cells for comparison of rear electrode performance. All ST-PSC cells reported have the same ITO front electrode. The ITO was deposited in Ar/O 2 atmosphere at the process pressure of 2 mTorr and oxygen to total flow ratio r(O 2 ) = 0.23%. The RF power density was 0.7 W cm −2 . An ≈10 nm thick hole transport layer (HTL) made of poly(triarylamine), (PTAA) and an ≈500 nm thick Cs 0.15 (FA 0.85 )Pb(I 0.98 Br 0.02 ) 3 perovskite absorber layer (bandgap of 1.55 eV) were deposited by spin coating. Further details on fabrication of the perovskite precursor and HTL solutions are provided in Glowienka et al. [21] Electron transport layers (ETLs) consisting of ≈20 nm of C 60 and ≈8 nm of Bathocuproine (BCP) were subsequently deposited by thermal evaporation. The fabrication of ST-PSC was finished by rear TCO deposition at room temperature of either ITO (≈180 nm) by RF-sputtering at Solliance, or IZrO (≈130 nm) by PLD at University of Twente. The thickness of both TCOs was optimized to obtain a sheet resistance (R sh ) of 50 Ω◻ −1 . Reference opaque solar cells were finished with thermal evaporation of 100 nm Cu films as the rear electrode. In case of PLD depositions, samples were sealed in a glovebox environment during both sample exchanges (before and after TCO deposition). MgF 2 layers were thermally evaporated on glass to reduce reflection losses in the visible range.
ST-PSC were measured under AM1.5G spectrum in a glovebox from the glass side (superstrate configuration). A stainless steel mask was used to define the active area of 0.09 cm 2 . The light intensity was calibrated by a silicon reference cell. The J-V curves were measured using a Keithley 2400 at a scanning rate of 200 mV s −1 . The stabilized PCE was obtained by tracking the output power for 5 min. Figure 2 displays the transmittance and absorptance curves as a function of wavelength for the 100-nm IZrO thin films grown by PLD and RF-sputtering before and after annealing. The films exhibit similar properties: low absorptance in the measured spectral range (300-1500 nm) for as-deposited state which further drops below 10% after the annealing step. Decreased absorptance in the near infrared (NIR) for the sputtered films, is caused by suppressed free carrier absorption (FCA) due to the decrease of N e ( Table 1) after the heat treatment. In contrast, PLD-grown films demonstrate negligible differences in NIR transmittance after annealing, which is also in-line with only a slight increase in free carrier density after annealing (Table 1). A common observation, for PLD and sputtered films, is the widening of the band gap after annealing. Previous reports suggest that this is linked to the phase transition from an amorphous to a polycrystalline state, [22][23][24] as will be demonstrated in the next section. This structural transition also affects electrical properties of the films. For IZrO deposited by either, PLD or sputtering, the Hall mobility increases after the annealing step at 200 °C reaching >70 cm 2 V −1 s −1 which is more than twice the as-deposited value.

Zr-Doped In 2 O 3 : Optoelectronic Properties of PLD-Grown Films and Comparison with Films Grown by RF-Sputtering
While the increase in μ e is consistent for films deposited with both methods, the N e present different trends. Three possible dopants could influence N e in IZrO films, namely: Zr + atoms, atomic hydrogen, and oxygen vacancies. The N e of the as-deposited sputtered films is higher than the N e of the as-deposited PLD films (6.2 vs. 4.7 × 10 20 cm −3 ), which intuitively indicates that the sputtered films have a larger amount of oxygen vacancies or higher hydrogen content as compared to the PLD-grown www.advmattechnol.de films. The N e of the sputtered IZrO drops considerably after the heat treatment, which has been previously explained by the uptake of oxygen by the films from the air. [25,26] Another possible cause of decreased N e , as described by Koida et al., [27] is hydrogen effusion or hydrogen passivation of oxygen vacancies during solid phase crystallization. Even though IZrO films are deposited without intentionally introducing water (like normally done for IO:H), H 2 O will be present at a background pressure of 10 −6 to 10 −7 mbar. Contrary, the PLD films have lower carrier density as-deposited, hinting toward more oxidized conditions (and/or leaner hydrogen content) and therefore a competition of oxygen/hydrogen uptake and desorption during annealing. The slight increase in N e suggests oxygen, hydrogen or H 2 O desorption. However, decoupling these causes is challenging and requires dedicated measurements. [27]

Tracking the Amorphous-to-Polycrystalline Transition in Zr-Doped In 2 O 3 Films Grown by PLD and RF-Sputtering
To our knowledge, solid phase crystallization of In 2 O 3 -based TCOs, widely reported for sputtered films, has been less reported for films grown by PLD. Moreover, the crystal orientation and growth mechanism of these grains have not been described in detail. Here we use GI-XRD and in situ heating to determine the preferential orientation of the films and the evolution of the crystallization. Figure 3 shows the XRD pattern of 100-nm asdeposited and annealed IZrO films. The absence of diffraction peaks for the as-deposited samples is indicative of mostly amorphous films. However, utilizing high-resolution GI-XRD ( Figure S1, Supporting Information), we additionally observe the formation of two very low-intensity peaks close to the characteristic crystallographic planes of the In 2 O 3 bixbyite cubic structure: (222) for the PLD-grown film and (400) plane for the sputtered film. Ex situ annealing of the samples at 200 °C leads to the formation of polycrystalline films with the initial preferential orientation of the nanocrystallites (Figure 3). GI-XRD scans during in situ heating in ambient atmosphere were furthermore performed at the temperature range from 125 to 225 °C to track the evolution of the crystallization and determine the amorphousto-polycrystalline transition temperature. The measurements at each temperature (2 min 40 s in duration) was preceded by 2 min 30 s of stabilization time. The temperature step size was 5 °C, the heating rate was fixed at 25 °C min −1 . The contour plot comprising the XRD patterns measured at each temperature step is presented in Figure 4.
There are two main observations from Figure 4: one is related to the threshold temperature for crystallization, while the other one is related to the distinct intensities of the main peaks. The threshold temperature for full crystallization (measured at the highest peak intensity) for the PLD films is systematically lower: 145 °C versus 155 °C for the sputtered samples. This indicates higher activation energy for the crystallization process of the sputtered films. An annealing in the vicinity of the transition temperature (150 °C) reveals different crystallization rates   , where X c is the crystalline fraction, K is a temperature-dependent rate constant and n is an exponent that reflects the geometry of the transformation. [28] The exponents n of the fitted curves were found to be ≈1.4 for sputtered and ≈2.2 for PLD-grown films ( Figure S3, Supporting Information). The former reflects site-saturated growth, while the latter is explained by continuous nucleation growth model as suggested by Wang et al. [29] We hypothesized that higher intensity of the main (222) peak in XRD scan for PLD IZrO film (Figure 4 and Figure S4, Supporting Information) and slightly bigger grains in the annealed state are due to the faster nucleation rate of these films. This was furthermore indicated by the larger value of Avrami exponent and higher density of nanocrystallites for PLD-grown films as shown by phase fraction determination analysis based on EBSD measurements (Table S1, Supporting  Information).
Additionally, the slight differences in the crystalline transition might be related to the stoichiometry of the films. Wang et al [29] studied the crystallization of In 2 O 3 in the presence and absence of water vapor and demonstrated the dependence of crystallization kinetics on the oxygen content of the film. As suggested by higher free carrier densities, sputtered films fabricated at lower process pressure and oxygen partial pressures than PLD films, may have higher amount of oxygen vacancies in the as-deposited state (Section 4.1). Other reports suggest a link between oxygen content and the preferential orientation of ITO, indicating that oxygen-deficient ITO present (200) preferential orientation, while stoichiometric films present (111) preferential orientation. [30][31][32] Nevertheless, there is no striking difference on the optoelectronic properties of the PLD and sputtered IZrO films (likely due to their polycrystalline nature) despite the distinct preferential orientation observed in all the experiments: (111) orientation in the case of PLD-grown films and (200) orientation in the case of sputtered films. This indicates that the optoelectronic properties of polycrystalline IZrO are independent of the preferential film orientation which, in its turn, is not a characteristic feature of the deposition method, but of the process parameters.

Electron Backscatter Diffraction
Electron backscatter diffraction (EBSD) was used to locally probe the crystallographic orientation of the IZrO films grown by PLD and by RF-sputtering. Figure 5 shows the EBSD mapping on IZrO films as-deposited (left) and annealed at 200 °C in air (right). The figures demonstrate the presence of mainly amorphous phase at the as-deposited state (shown as black areas) with the presence of nanocrystallites with In 2 O 3 phase. It is important to mention that pole figures represent crystal directions or plane normal of a material within the sample reference system. For cubic materials it is reduced to the standard triangle with (001), (111) and (010) orientations. Interestingly, the sputtered films show a higher density of nanocrystallites parallel to (001) and (010) orientations, while the nanocrystallites of the PLD-grown are mainly parallel to (111) direction. This preferential orientation of the peak is preserved after annealing, as observed by the higher density of (111) oriented grains (green areas) in the annealed PLD-grown films. Indepth analysis of the grain orientations at the X, Y, and Z axis is shown in Figure S5, Supporting Information. The EBSD results confirm the XRD data and, moreover, indicate that the preferential orientation of the nanocrystallites, formed during the deposition of the films, defines the final film orientation. Figure 5 and scanning electron microscope (SEM) images ( Figure S6, Supporting Information) of annealed films indicate grain sizes of 200-300 nm in width, for films grown with both methods. PLD IZrO films seem to have slightly larger grains as also suggested by the higher intensity of the main peak in XRD measurements (Figure 3). This could be a consequence of the lower density of crystalline seeds in the PLD as-deposited IZrO, leading to slightly more space for the grains to grow as compared to the sputtered IZrO. This is in line with previous reports describing the effect of the density of crystalline seeds with the final grain size in IZrO. [23] The correlation of density of crystalline seeds and final grain size was also reported for Wand Ce-doped In 2 O 3 films deposited via sputtering and reactive plasma deposition (RPD). [24] The formation of large and high crystal quality grains has a direct effect enhancing the electron mobility measured for both films and enable fabrication of thin films with comparable μ e despite slight difference in crystal size. [24,33] The preferential orientation of these films, in its turn, www.advmattechnol.de appears to have minimal effect on the optoelectronic properties, as mentioned above.

Integration of PLD-Grown IZrO in Semi-Transparent Halide Perovskite Solar Cells (ST-PSC)
ST-PSC intended for both mechanically stacked four-terminal (4T) or monolithically integrated two-terminal (2T) tandem devices have two similar main bottlenecks to overcome related to transparent electrodes. [34] One of them is loss of NIR transmittance due to parasitic absorption in the TCOs [35] which is more pronounced for 4T case due to the increased number of transparent electrodes. The second one is integration of the TCO electrode by physical vapor deposition (PVD) methods on top of the sensitive halide perovskite and contact layers. The former is typically solved by utilizing TCOs with high mobility, [36] while the latter is approached by inserting additional solution-processed metal oxide buffer layers [37] or using "soft" landing methods on top of the perovskite layer. [17,38] Our main objective is to show that PLD may serve as an alternative for damage-free deposition of TCOs in solar cells. To prove the applicability of PLD for scalable fabrication of efficient ST-PSC for 4T devices, we implemented IZrO (as-deposited, mostly amorphous state) as the rear electrode and compared cell performance with reference opaque cells and semi-transparent cells with sputtered rear ITO. For a fair comparison of the electrical performance of the cells, the thicknesses of the IZrO (≈130 nm) and ITO (≈180 nm) were optimized to achieve R sh 50 Ω◻ −1 (both deposited at room temperature without posttreatment). A comparison of optoelectrical properties for typical TCO films employed in this study are provided in Figure S7 and Table S3, Supporting Information. The device stack and the different rear electrodes are shown in Figure 6a. Figure 6b displays statistical distribution of power conversion efficiencies (PCE) for solar cells with different rear electrodes for both forward and reverse scan directions measured from the glass side. The star indicates the stabilized PCE that was obtained by tracking the output power for 300 s as shown in Figure 6c. No efficiency deterioration is observed for cells with either of the rear electrodes. PCE of the reference opaque cell is reaching 19.8% for the champion cell highlighting high quality of the halide absorber and relevant choice of the transport layers. Stabilized PCE of the ST-PSC cells with PLD-grown  www.advmattechnol.de rear TCO is higher than of the counterparts with sputtered rear ITO (15.1% vs. 11.9%).
As seen in Figure 6b, cells with IZrO electrode demonstrate small variations on PCE. This proves batch-to-batch reproducibility in addition to uniformity of properties within a single deposition, as four substrates (9 cm 2 area), with 4 cells each, were simultaneously loaded on a 4-inch carrier wafer (Figure 7a). Further insights into the origins of the performance difference for ST-PSC could be extracted from the box plot of the solar cell parameters in Figure 7b-d: short-circuit current density (J sc ), open-circuit voltage (V oc ) and fill factor (FF). The J sc of cells with either of the transparent electrodes demonstrate comparable values of around 21 mA cm −2 . This can be explained by the identical front ITO electrode leading to similar transparency in the spectral region of strong halide perovskite absorption (300-800 nm). However, the wider bandgap of IZrO as compared to ITO ( Figure S7, Supporting Information) may lead to the increased photocurrent from UV photons if IZrO is placed also as front electrode in ST-PSC. Noteworthy, high transmittance in the 800-1200 nm region for both IZrO and ITO in the study is beneficial to increase J sc in tandem devices with the bottom cell harvesting the NIR part of the spectrum. [20,35,39] In contrast to the J sc , the FF, and V oc differ significantly for ST-PSC with different rear TCOs. Cells with IZrO electrode demonstrate higher FF (68% vs. 55%) and V oc (1050 mV vs. 1000 mV). Although there is some inhomogenity in the cell parameters for ST-PSC with ITO and evaporated ETL in this experiment, we believe the results to be indicative of the performance deterioration for the cells with this specific sputtered ITO. First, it was previously shown [39,40] that usage of 30-nm thick compact ZnO deposited by spatial atomic layer deposition on top of ETL is critical to protect the cell during the described sputtering process of the top ITO layers. In this contribution we purposefully omit the additional layer to prove the advantage of PLD mitigating as a damage-free deposition process in contrast to sputtering. Second, the observed S-shaped J-V degradation ( Figure S8, Supporting Information) for cells with ITO electrode reveals the sputter-induced damage of organic ETLs for the investigated devices. This potentially leads to the increased barrier height at ETL/TCO interface as suggested by Kanda et al. [41] causing severe drop of FF and V oc . This J-V degradation is not pronounced for the cells with PLD grown rear electrode. We link this to the higher deposition pressure during PLD in this study enabling thermalization of species arriving on the solar cell precursors. [5,18] These process conditions could be achieved with other deposition methods including sputtering, [17] however ignition at low power (a common requirement for low-damage sputter deposition) and high pressures may be problematic. [42] Moreover, optoelectronic properties of TCO films and their density may be compromised at these process conditions. [43,44] Although the demonstrated "soft" landing deposition is likely linked to the process conditions utilized specifically for the compared cases, PLD as a method is favoring larger deposition pressure range, since the energy source for material ablation (laser) is physically decoupled from the vacuum equipment. Additional experiments with sputtered films at high pressures (matching those of PLD) will help elucidate if other features from the methods influence the damage on the substrates. For example, more confined plasma during PLD (as compared to sputtering) may be beneficial for reducing the negative ion flux [45,46] and thus mitigating the plasma damage.

www.advmattechnol.de
Further studies for TCOs deposited by PLD are required to understand the relation between the process parameters and the properties of TCO/transport layer interface and, subsequently, solar cell performance. Nevertheless, obtained results clearly demonstrate that PLD holds a great potential for the scalable fabrication of high-efficiency solar cells enabling deposition on top of sensitive carrier transport layers as in case of perovskite solar cells.

Conclusion
In summary, we have developed a room-temperature pulsed laser deposition process to obtain Zr-doped In 2 O 3 thin films with excellent optoelectronic properties on 4-inch wafers. High mobility (>60 cm 2 V −1 s −1 ) and low absorptance (<10%) in 300-1200 nm spectral range after a mild annealing step at 200 °C in air makes PLD-grown electrodes appealing for silicon/perovskite four-terminal tandem solar cells. Higher deposition pressure during ablation (as compared to sputtering) enables damage-free TCO deposition on top of sensitive materials. High stabilized efficiency (15.1%) of semi-transparent perovskite solar cell with IZrO electrode produced by PLD verifies the great potential for this vacuum method for scalable fabrication of optoelectronic devices.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.