Electrodeposited PEDOT:PSS-Al2O3 Improves the Steady-State Efficiency of Inverted Perovskite Solar Cells

The atomic layer deposition (ALD) of Al2O3 between perovskite and the hole transporting material (HTM) PEDOT:PSS has previously been shown to improve the efficiency of perovskite solar cells. However, the costs associated with this technique make it unaffordable. In this work, the deposition of an organic–inorganic PEDOT:PSS-Cl-Al2O3 bilayer is performed by a simple electrochemical technique with a final annealing step, and the performance of this material as HTM in inverted perovskite solar cells is studied. It was found that this material (PEDOT:PSS-Al2O3) improves the solar cell performance by the same mechanisms as Al2O3 obtained by ALD: formation of an additional energy barrier, perovskite passivation, and increase in the open-circuit voltage (Voc) due to suppressed recombination. As a result, the incorporation of the electrochemical Al2O3 increased the cell efficiency from 12.1% to 14.3%. Remarkably, this material led to higher steady-state power conversion efficiency, improving a recurring problem in solar cells.


Introduction
The solution processing of organic-inorganic perovskite solar cells is a promising route for the fabrication of cost-effective photovoltaic devices [1,2]. These solar cells can be grouped into two main architectures: direct and inverted. They differ in the direction of electron flow within the device, flowing towards the conductive glass for the direct devices and towards the metal electrode for the inverted architecture. This requires the use of different charge extraction materials. In direct architecture devices, the light is incident on the conductive oxide and travels across a transparent electron-transporting material (ETM) before reaching the perovskite. For this architecture, the most commonly used materials are TiO 2 as ETM, and Spiro-OMeTAD as HTM. However, high costs, hightemperature processes (450 • C), and hysteresis in cells made with these materials have led to the development of the inverted architecture. Unlike direct solar cells, inverted architecture features devices with the HTM being deposited over the conductive oxide, hence this material must be transparent to allow incident light to reach the active layer [3,4]. Research on HTMs for perovskite solar cells is currently very relevant, as important requirements have not yet been fulfilled. High hole mobility is needed to ensure fast charge transport and high chemical and electrical stability, and conducting this at a low cost is also necessary [5][6][7].
PEDOT:PSS is a conductive polymer widely used as HTM in inverted perovskite solar cells due to its high transparency, flexibility, and low commercial cost. One of the most significant benefits of this polymer is the variety of techniques available for its deposition, as it can be spin-coated, spray-coated, doctor-blade coated, and even electrodeposition coated [8]. Nevertheless, achieving high power conversion efficiency (PCE) PEDOT:PSS-Cl was performed in a three-electrode cell; the working electrode was an ITO substrate; a platinum wire was used as a counter-electrode, and Ag/AgCl (3 M NaCl) as the reference electrode. The synthesis solution contained 4.5 g/L NaPSS (Aldrich, Average Mw 70,000, St. Louis, MO, USA), 0.02% v/v EDOT (Aldrich, 97%, St. Louis, MO, USA), and 0.1 M NaCl (Panreac ≥ 99.5%, Darmstadt, Germany). This solution was bubbled with N 2 for 5 min and electropolymerization was conducted by cyclic voltammetry between −0.1 and 1.2 V at 100 mVS −1 until a charge of 4 mC/cm 2 was reached.

Electrodeposition of Al 2 O 3 on PEDOT:PSS-Cl
Obtaining Al 2 O 3 by electrodeposition requires thermal annealing to achieve the transition from Al(OH) 3 [23,24]. A lower thickness facilitates the transition and on the nanometer scale, and it starts as low as 200 • C [25,26]. Thus, to obtain the PEDOT-Al bilayer, PEDOT:PSS-Cl substrates were used as working electrodes to which a potential of −1 V (vs Ag/AgCl) was applied in a 0.02 M AlCl 3 aqueous solution (pH 4.2). As a result, an Al(OH) 3 deposit was formed and the substrates were thermally annealed at 200 • C, 225 • C, and 240 • C for 1 h. To obtain nanometer scale deposits, the electrodeposition charge was set at 0.50 mC cm −2 and 0.75 mC cm −2 .

Solar Cell Fabrication
To test the effect of Al 2 O 3 on the performance of PEDOT:PSS-Cl as HTM, inverted perovskite solar cells were fabricated in a glovebox with a constant N 2 flow. First, the ITO/PEDOT:PSS-Cl substrates were annealed at 110 • C for 10 min, then 30 µL of perovskite precursor solution were deposited. This solution was prepared by dissolving 1.2 mmol of PbI 2 (99,9985%, Alfa Aesar, Haverhill, MA, USA) and 1.2 mmol of methylammonium iodide (Greatcellsolar, Queanbeyan, Australia) in 1 mL of a mixed solvent containing DFM and DMSO in a volumetric ratio of 10:1. Spin coating was started with a two-stage program, 10 s at 1000 rpm followed by 15 s at 5000 rpm. After 6 s of starting the second stage, 350 µL of chlorobenzene were added, and then the PCBM layer, which acts as ETM, was deposited by dynamic spin coating (25 µL of a 20 mg/mL solution in chlorobenzene at 5000 rpm for 30 s). Subsequently, the substrate was heated at 100 • C for 30 min, which is known as merged annealing [27]; the obtained perovskite layer was about 400 nm thick. Finally, the BCP layer was spin-coated from 40 µL of a 0.5 mg/mL solution in methanol:toluene 100:1 at 4000 rpm for 40 s. The PCBM/BCP layer was about 25 nm, the BCP helps to form an ohmic contact between PCBM and Ag, preventing direct contact between both materials [28]. The device was removed from the glovebox and taken to a thermal evaporator in which 100 nm of Ag were deposited by monitoring its thickness using a quartz sensor; the first 5 nm were deposited at a rate of 0.2 Å s −1 , and, thereafter, the rate was kept at 1 Å s −1 .

Characterization Methods
The solar cells were characterized with JV curves using AM 1.5G light (100 mWcm −2 ) with a solar simulator (Abet Technologies model 10500, Milford, CT, USA) in a potential range of −0.1 to 1.1 V at 50 mVs −1 (Autolab µ3AUT71099, Utrecht, The Netherlands). The light intensity was calibrated with a Hamamatsu S1133 photodiode, and the illuminated area of the devices was defined with a 0.065 cm 2 black shadow mask. The steady-state efficiency was determined with an unencapsulated cell; first its efficiency (PCE) was calculated from the JV curve, then a constant voltage was found and applied to keep the efficiency at a maximum. The steady-state PCE was monitored over time under a N 2 atmosphere. For stability during impedance spectroscopy (IS) measurements under a N 2 atmosphere, the devices were encapsulated with hot glue [29]. An illumination of 1 sun with an AC signal of 0.075 V was used, and measurements were made up to 0.6 V DC [30]. UV-Vis spectra were taken on an Analytik Jena SPECORD 50 PLUS spectrophotometer (Jena, Germany) and PL measurements on a Cary Varian Eclipse Fluorescence Spectrophotometer with an excitation wavelength of 405 nm. Impedance measurements and electrochemical characterizations were performed on an Autolab AUT84194 poten- tiostat equipped with the FRA module. HOMO level estimation by cyclic voltammetry was performed in anhydrous acetonitrile (Sigma Aldrich, HPLC grade, Darmstadt, Germany) with 0.1 M TBAPF6 (Aldrich, 98%, St. Louis, MO, USA) at 100 mV s −1 , using 1 mM ferrocene (Alfa Aesar, 99%, Haverhill, MA, USA) as a vacuum electrochemical reference. The potential was measured against an Ag/AgCl reference electrode using a double-junction chamber filled with 3 M of NaCl. Mott-Schottky data were taken by applying a 10 mV AC perturbation at 1000 Hz, and each applied DC potential point was held 10 s before taking the measurement. Electrochemical impedance spectroscopy (EIS) was performed at 205 mV DC (vs. Ag/AgCl) over a frequency range of 100 kHz to 0.1 Hz with a 10 mV AC perturbation. Raman spectra were obtained with an Xplora Horiba Scientific confocal Raman microscope (Kyoto, Japan) employing a 532 nm laser, and a height scan was performed to obtain a clear spectrum of the films. AFM images were taken with an Asylum Research microscope model MFP-3D-BIO in tapping mode. SEM images were taken with a Tescan Lyra 3 microscope (Brno, Czech Republic). Finally, contact angles were measured using an Attension theta model contact angle meter.

Influence of PEDOT-Al Bilayer on the Performance of Inverted Perovskite Cells
The photovoltaic parameters of perovskite solar cells employing different PEDOT-Al layers as HTM were compared with the PEDOT:PSS-Cl reference device. To find suitable fabrication conditions for the Al 2 O 3 layer, the annealing temperatures (200 • C, 225 • C, and 240 • C) and the electrodeposition charges per square centimeter (0.50 mC and 0.75 mC) were varied. The resulting photovoltaic parameters for each variation are illustrated in Figure 1 and summarized in Table S1. Improvements in power conversion efficiency (PCE), short current density (J sc ), and open-circuit voltage (V oc ) were observed for all cells in which the PEDOT-Al bilayer was used. It was found that the electrodeposition charge has no significant effect on the studied parameters, but the efficiency and fill factor (FF) of the cells decreased with increasing annealing temperature. Based on these results, it was determined that the best electrodeposition conditions for Al 2 O 3 are a 200 • C annealing temperature and 0.50 mC charge. The latter was chosen for simplicity, as it optimizes the deposition time (around 6 s).
UV-Vis spectra were taken on an Analytik Jena SPECORD 50 PLUS spectrophotometer (Jena, Germany) and PL measurements on a Cary Varian Eclipse Fluorescence Spectrophotometer with an excitation wavelength of 405 nm. Impedance measurements and electrochemical characterizations were performed on an Autolab AUT84194 potentiostat equipped with the FRA module. HOMO level estimation by cyclic voltammetry was performed in anhydrous acetonitrile (Sigma Aldrich, HPLC grade, Darmstadt, Germany) with 0.1 M TBAPF6 (Aldrich, 98%, St. Louis, MO, USA) at 100 mV s −1 , using 1 mM ferrocene (Alfa Aesar, 99%, Haverhill, MA, USA) as a vacuum electrochemical reference. The potential was measured against an Ag/AgCl reference electrode using a double-junction chamber filled with 3 M of NaCl. Mott-Schottky data were taken by applying a 10 mV AC perturbation at 1000 Hz, and each applied DC potential point was held 10 s before taking the measurement. Electrochemical impedance spectroscopy (EIS) was performed at 205 mV DC (vs Ag/AgCl) over a frequency range of 100 kHz to 0.1 Hz with a 10 mV AC perturbation. Raman spectra were obtained with an Xplora Horiba Scientific confocal Raman microscope (Kyoto, Japan) employing a 532 nm laser, and a height scan was performed to obtain a clear spectrum of the films. AFM images were taken with an Asylum Research microscope model MFP-3D-BIO in tapping mode. SEM images were taken with a Tescan Lyra 3 microscope (Brno, Czech Republic). Finally, contact angles were measured using an Attension theta model contact angle meter.

Influence of PEDOT-Al Bilayer on the Performance of Inverted Perovskite Cells
The photovoltaic parameters of perovskite solar cells employing different PEDOT-Al layers as HTM were compared with the PEDOT:PSS-Cl reference device. To find suitable fabrication conditions for the Al2O3 layer, the annealing temperatures (200 °C, 225 °C, and 240 °C) and the electrodeposition charges per square centimeter (0.50 mC and 0.75 mC) were varied. The resulting photovoltaic parameters for each variation are illustrated in Figure 1 and summarized in Table S1. Improvements in power conversion efficiency (PCE), short current density (Jsc), and open-circuit voltage (Voc) were observed for all cells in which the PEDOT-Al bilayer was used. It was found that the electrodeposition charge has no significant effect on the studied parameters, but the efficiency and fill factor (FF) of the cells decreased with increasing annealing temperature. Based on these results, it was determined that the best electrodeposition conditions for Al2O3 are a 200 °C annealing temperature and 0.50 mC charge. The latter was chosen for simplicity, as it optimizes the deposition time (around 6 s).  Figure 2a). These results are lower than the 20.22% PCE reported for spin-coated CsI-PEDOT:PSS devices [31], but represent an improvement in the performance of electrodeposited PEDOT-based devices. As mentioned in the Introduction, the best PCE reported to date using an electrodeposited PEDOT-based MTH exhibits 13.56% (J sc : 22.19 mA cm −2 , V oc : 0.94 V, and FF: 65%) [22]. In comparison, the results reported in this study correspond to a significantly higher FF, which compensates for the lower J sc , resulting in a higher PCE. Previously, the increased efficiency of solar cells using PEDOT/Al 2 O 3 has been attributed to the improved charge selectivity [18]. Figure 2b shows the architecture used in this study, in which PEDOT has an energy configuration with a density of states (DOS) appropriate for hole extraction and unfavorable for electron transport. Therefore, holes have a higher probability of tunneling through the thin insulating layer than electrons ( Figure 2c) [16,17].
ing PEDOT-Al, only the variations are displayed on the labels. The red squares represent the mean and the orange lines the median (10 devices each variation).
The reference solar cells reached a maximum efficiency of 12.1% (Jsc: 17.5 mA cm −2 , Voc: 0.905 V, and FF: 77%). In contrast, solar cells employing PEDOT-Al (0.50 mC, 200 °C) exhibited a maximum efficiency of 14.3% (Jsc: 19.9 mA cm −2 , Voc: 0.956 V, and FF: 75%) (Figure 2a). These results are lower than the 20.22% PCE reported for spin-coated CsI-PEDOT:PSS devices [31], but represent an improvement in the performance of electrodeposited PEDOT-based devices. As mentioned in the Introduction, the best PCE reported to date using an electrodeposited PEDOT-based MTH exhibits 13.56% (Jsc: 22.19 mA cm −2 , Voc: 0.94 V, and FF: 65%) [22]. In comparison, the results reported in this study correspond to a significantly higher FF, which compensates for the lower Jsc, resulting in a higher PCE. Previously, the increased efficiency of solar cells using PEDOT/Al2O3 has been attributed to the improved charge selectivity [18]. Figure 2b shows the architecture used in this study, in which PEDOT has an energy configuration with a density of states (DOS) appropriate for hole extraction and unfavorable for electron transport. Therefore, holes have a higher probability of tunneling through the thin insulating layer than electrons ( Figure  2c) [16,17]. A recurrent issue in perovskite solar cells is that the PCE measured from the JV curves (JV PCE) is not the actual steady-state power conversion efficiency maintained by the device [32]. Discrepancies between the PCE from the JV and the steady-state PCE are related to hysteresis in the JVs as well as to slow transient phenomena in the perovskite A recurrent issue in perovskite solar cells is that the PCE measured from the JV curves (JV PCE) is not the actual steady-state power conversion efficiency maintained by the device [32]. Discrepancies between the PCE from the JV and the steady-state PCE are related to hysteresis in the JVs as well as to slow transient phenomena in the perovskite devices. The latter is connected to the ionic migration in the perovskite film, which causes changes in the internal electric field, charge recombination rates, and the interfacial trap states [33,34].
To measure the steady-state PCE, a constant potential of 0.75 V was applied to the cells for 45 min (see Figure 2d-f). When analyzing the first 200 s of the test, the PEDOT-Al 0.50 mC 200 • C cells demonstrated a slightly higher stabilized efficiency compared to their JV PCE, whereas PEDOT:PSS-Cl devices with an initial JV PCE of 12.3% were only able to maintain up to 92% of their JV PCE. Moreover, during the entire test, the steady-state PCE of the reference cell decreased by 4.2%, while the PEDOT-Al cell only had a loss of 2.2%. The trend of the curves in Figure 2f illustrates the significance of this result and the effect that the Al 2 O 3 layer could have in the long term performance of the cells under continuous operation. The cells with PEDOT-Al also showed better shelf stability; after 25 days, the PEDOT:PSS-Cl devices maintained 75% of their initial FF, while the PEDOT-Al devices maintained 90% ( Figure S1). The above meant that, for the PEDOT:PSS-Cl reference device, an average of 70% of its initial efficiency was maintained, compared to 77% for the PEDOT-Al devices. This also represents an advantage of PEDOT electrodeposition, since spin-coated PEDOT:PSS devices tend to degrade completely in about 20 days of storage [20,22].
A major drawback to obtain stable perovskite solar cells is the reactivity of the PEDOT:PSS/perovskite interface. First, it has been shown that methylammonium ions (MA + ) can interact with PSS − causing the PEDOT:PSS work function to decrease [35]. These ions tend to accumulate at this interface under dark conditions due to the built-in potential (V bi ) of the solar cell [36]. Second, it has recently been observed that halide ions (I − , Cl − , and Br − ) in the perovskite precursor solution dope the PEDOT:PSS [37], introducing mid-gap states leading to recombination [14,20]. Under illumination, near the maximum power point (V mpp ) the net potential is reversed and I − ions accumulate at the PEDOT:PSS/perovskite interface [36]. To test whether the Al 2 O 3 layer could mitigate these deleterious effects on the PEDOT layer, a Mott-Schottky analysis of the HTMs in the presence of halide ions (Cl − ) was performed. The tests were conducted by polarizing the polymer to provoke doping/dedoping processes.
The doping/dedoping effects induced by halide ions in PEDOT:PSS-Cl films are demonstrated by the Mott-Schottky plots ( Figure S2a,b). The reverse scans (taken from 0.7 to −0.5 V) show a negative slope around −0.25 V, which is consistent with the p-doping of the polymer [38]. However, the negative slope was not observed after changing the scan direction (forward). To rule out that this trend was a product of polymer degradation, a third reverse scan was performed in which the slope reappeared, suggesting reversible behavior; the results were reproducible in several replicates. Consequently, this behavior can be attributed to the doping/dedoping process of the polymer. As for the PEDOT-Al bilayers, Mott-Schottky plots in Figure S2c-f for both scan directions show the expected negative slope for p-type polymers. This suggests that the electrodeposited inorganic layer, regardless of its deposition parameters, protects the polymer from being doped/dedoped. This agrees with the results obtained for steady-state PCE: a more constant steady-state efficiency for PEDOT-Al devices, which are protected from doping/dedoping effects, and an efficiency that decreases rapidly with time for reference cells, where the polymer is exposed to I-ions that induce changes in its electrical properties (Figure 2f) [36].
The effect of PEDOT-Al 0.50 mC annealing temperature on perovskite film morphology was studied through top-view SEM images ( Figure 3). Grains up to~1.5 µm were found in all variations, possibly due to the merged annealing method that produces large-grain-size perovskite films [27]. Furthermore, when analyzing the average grain size, no trend was detected to explain the higher efficiencies with PEDOT-Al. The average grain size for annealing at 200 • C is even smaller than that of the reference device. This shows that the reason for the efficiency increase is not related to the perovskite grain size [39]. It can also be observed that the grain size distribution is above the perovskite layer thickness (400 nm) for all studied devices ( Figure S3). Variations within these ranges do not significantly increase the grain boundaries in the vertical direction and do not induce larger recombination sites.
(400 nm) for all studied devices ( Figure S3). Variations within these ranges do not significantly increase the grain boundaries in the vertical direction and do not induce larger recombination sites.  [40]. In addition, it was observed that the annealing temperature of 240 °C does not affect the thickness of PEDOT-Al 0.50 mC, thus polymer shrinkage due to thermal degradation was discarded.
The morphology of electrodeposited bilayers was compared with that of bilayers obtained by physical vapor deposition (PVD). For this purpose, a ~2 nm Al layer was thermally evaporated on PEDOT:PSS-Cl and annealed at 200° C. The thickness (16-18 nm) and homogeneity of the resulting PVD bilayer suggest that the electrochemical process offers similar quality and film thickness control as PVD. Regarding the perovskite layer, cross-sectional SEM images show similar monolithic vertical grains for all variations (Figure S3), which is beneficial for cell performance and stability [41].
By AFM imaging, it was observed that the ITO substrate has a root mean square (RMS) roughness of 1.7 nm ( Figure S5). After electrodepositing the HTMs, the RMS roughness increased by a similar value for all variations (around 3 nm), even for the inorganic layer by PVD. This indicates that the homogeneity of the electrodeposited Al2O3 is comparable to that obtained by PVD. An important advantage of the electrodeposition of polymers and oxides is the low film thickness and roughness that can be obtained [22,40,42]. As for hydrophilicity, contact angle measurements showed that Al2O3 has a hydrophobic  [40]. In addition, it was observed that the annealing temperature of 240 • C does not affect the thickness of PEDOT-Al 0.50 mC, thus polymer shrinkage due to thermal degradation was discarded.
The morphology of electrodeposited bilayers was compared with that of bilayers obtained by physical vapor deposition (PVD). For this purpose, a~2 nm Al layer was thermally evaporated on PEDOT:PSS-Cl and annealed at 200 • C. The thickness (16-18 nm) and homogeneity of the resulting PVD bilayer suggest that the electrochemical process offers similar quality and film thickness control as PVD. Regarding the perovskite layer, cross-sectional SEM images show similar monolithic vertical grains for all variations ( Figure S3), which is beneficial for cell performance and stability [41].
By AFM imaging, it was observed that the ITO substrate has a root mean square (RMS) roughness of 1.7 nm ( Figure S5). After electrodepositing the HTMs, the RMS roughness increased by a similar value for all variations (around 3 nm), even for the inorganic layer by PVD. This indicates that the homogeneity of the electrodeposited Al 2 O 3 is comparable to that obtained by PVD. An important advantage of the electrodeposition of polymers and oxides is the low film thickness and roughness that can be obtained [22,40,42]. As for hydrophilicity, contact angle measurements showed that Al 2 O 3 has a hydrophobic nature ( Figure S6a-f). Hence, when used to modify PEDOT:PSS-Cl films, it causes the polymer to become more hydrophobic. This could explain the slight improvement in the shelf stability of the unencapsulated PEDOT-Al devices compared to PEDOT:PSS-Cl.

Spectroscopic and Electrochemical Characterization
Additional spectroscopic characterizations were performed to identify the reasons why PEDOT-Al is a better HTM than PEDOT:PSS-Cl.
Considering that the perovskite layer had the same thickness in all devices, it was confirmed by UV-visible absorption spectroscopy that perovskite films deposited on the different substrates show no differences in light absorption (Figure 4a). Stationary photoluminescence (PL) showed that all HTMs induce strong PL quenching compared to perovskite deposited on ITO (Figure 4b), and PEDOT:PSS-Cl showed the highest PL quenching. Although this seems to go against the higher J SC found for PEDOT-Al bilayers, it has been demonstrated that, when the most intense PL corresponds to the most efficient device, it is considered indicative that the HTM modification induces changes in recombination phenomena from a non-radiative to a radiative process [43,44]. In the PEDOT-Al bilayers, Al 2 O 3 could produce this effect by passivating the non-coordinated Pb 2+ ions on the perovskite surface, thus reducing the trap states responsible for non-radiative recombination, which correlates with the V oc enhancement [18,[43][44][45]. nature ( Figure S6a-f). Hence, when used to modify PEDOT:PSS-Cl films, it causes the polymer to become more hydrophobic. This could explain the slight improvement in the shelf stability of the unencapsulated PEDOT-Al devices compared to PEDOT:PSS-Cl.

Spectroscopic and Electrochemical Characterization
Additional spectroscopic characterizations were performed to identify the reasons why PEDOT-Al is a better HTM than PEDOT:PSS-Cl.
Considering that the perovskite layer had the same thickness in all devices, it was confirmed by UV-visible absorption spectroscopy that perovskite films deposited on the different substrates show no differences in light absorption (Figure 4a). Stationary photoluminescence (PL) showed that all HTMs induce strong PL quenching compared to perovskite deposited on ITO (Figure 4b), and PEDOT:PSS-Cl showed the highest PL quenching. Although this seems to go against the higher JSC found for PEDOT-Al bilayers, it has been demonstrated that, when the most intense PL corresponds to the most efficient device, it is considered indicative that the HTM modification induces changes in recombination phenomena from a non-radiative to a radiative process [43,44]. In the PEDOT-Al bilayers, Al2O3 could produce this effect by passivating the non-coordinated Pb 2+ ions on the perovskite surface, thus reducing the trap states responsible for non-radiative recombination, which correlates with the Voc enhancement [18,[43][44][45]. As previously mentioned, one of the reasons for the improved performance of PE-DOT-Al devices could be the presence of an energy barrier that increases charge selectivity. In order to test this, an estimation of the HOMO level was performed by cyclic voltammetry. However, it is clarified that these values are used for comparative purposes between variations, since this characterization is carried out in liquid media, which is drastically different from the operating conditions of solar cells. Even so, this technique is widely used, as it correctly reproduces the trends of other accurate, but more expensive, photoemission-based techniques [46][47][48][49][50]. Ferrocene was used as a reference for the vacuum level, and the HOMO level was calculated using the onsets of the oxidation potentials ( Figure S7 and Table S2). The HOMO level values obtained were 5.06 eV for PE-DOT:PSS-Cl and 5.10 eV for all PEDOT-Al bilayers, demonstrating the additional energy barrier imposed by the inorganic layer. These results concur with previous works on ALDdeposited Al2O3. For example, it was reported that the Al2O3 layer on PEDOT-Al has a passivating effect and imposes an energy barrier that increases charge selectivity, which results in increased PCE, Voc, and Jsc [18,51]. Similarly, when ALD-Al2O3 is used as an intermediate layer between perovskite and Spiro-OMeTAD HTM, there is an increase in As previously mentioned, one of the reasons for the improved performance of PEDOT-Al devices could be the presence of an energy barrier that increases charge selectivity. In order to test this, an estimation of the HOMO level was performed by cyclic voltammetry. However, it is clarified that these values are used for comparative purposes between variations, since this characterization is carried out in liquid media, which is drastically different from the operating conditions of solar cells. Even so, this technique is widely used, as it correctly reproduces the trends of other accurate, but more expensive, photoemissionbased techniques [46][47][48][49][50]. Ferrocene was used as a reference for the vacuum level, and the HOMO level was calculated using the onsets of the oxidation potentials ( Figure S7 and Table S2). The HOMO level values obtained were 5.06 eV for PEDOT:PSS-Cl and 5.10 eV for all PEDOT-Al bilayers, demonstrating the additional energy barrier imposed by the inorganic layer. These results concur with previous works on ALD-deposited Al 2 O 3 . For example, it was reported that the Al 2 O 3 layer on PEDOT-Al has a passivating effect and imposes an energy barrier that increases charge selectivity, which results in increased PCE, V oc , and J sc [18,51]. Similarly, when ALD-Al 2 O 3 is used as an intermediate layer between perovskite and Spiro-OMeTAD HTM, there is an increase in PCE from 15.1% to 18.0%. The ability to use ALD on any surface is a remarkable advantage, as it allows Al 2 O 3 to be deposited in direct structure devices in which the hydrophobic nature of this layer is an effective barrier between the ambient humidity and the perovskite layer. As a result, the device stability is notoriously increased, whereby solar cells have been reported to maintain 60-70% of their original PCE even after 70 days of storage [17,52]. e-Beam processed interlayer materials between the HTM and the perovskite have also showed similar improvements to those reported here. On MAPbBr 3 perovskite devices, the HOMO level of NiOx HTM have been improved by using an overlayer of e-Beam MoOx. The efficiency goes from 2.79% for NiOx to 5.2% for the NiOx/e-Beam MoOx solar cells. The MAPbBr 3 perovskite shows relatively low PCE nonetheless it shows promising high V oc values, which can be useful for applications such as solar driven water electrolysis, photocatalysis, and multijunction solar cells. A remarkable 1.653 V V oc (10.08% PCE) has been achieved by using interlayers at both sides of the perovskite, e-Beam MoOx and ALD ZrO 2 for the HTM and ETM interfacial modification [53].
To understand the noticeable decrease in FF with the annealing temperature of 240 • C (Figure 1d), the series (R s ) and shunt resistances (R sh ) of the solar cells were plotted (Figure 5a,b). For optimum FF, the R s should be as low as possible and R sh should be high. Nevertheless, it has been observed that, in perovskite solar cells, the FF and therefore the PCE are particularly affected by R s [54], and the influence of shunt resistance (R sh ) is only noticeable below 1000 Ω cm 2 . Ahmed et al. found similar results, where a R s as low as 6 Ω cm 2 can result in a poor FF (67%) [55]. Therefore, the decrease in FF for the annealing temperature of 240 • C can be attributed to the increase in R s . This resistance can be caused by ohmic elements, such as metallic contact, ETM, HTM, or conductive substrate [54]. Since the HTM is the layer that is varying, the observed changes in R s were assigned to the R s of this material. Moreover, the annealing process could induce thermal degradation of PEDOT, which decreases its conductivity [56].  Raman spectroscopy characterization of PEDOT:PSS-Cl and PEDOT-Al 0.50 mC was performed in search of thermal degradation signals that would explain the increase in R s (Figure 5c). All the samples analyzed showed the characteristic PEDOT:PSS signals at 440 cm −1 , 575 cm −1 , 990 cm −1 , 1252 cm −1 , 1366 cm −1 , 1445 cm −1 , 1500 cm −1 , and 1570 cm −1 [57,58]. Thus, no apparent degradation was found even at the annealing temperature of 240 • C. Nonetheless, the relative Raman intensities were compared and it was noticed that the signals around 650 cm −1 , 680 cm −1 , and between 1000 cm −1 and 1200 cm −1 increased with annealing temperature (insets Figure 5c). The increase in these signals has been related to low-chain planarity, and structural defects in the polymeric chains due to the formation of side groups during overoxidation [59,60]. The degradation of PEDOT by overoxidation has been linked to photo-oxidation, electrochemical overoxidation, and thermal degradation. These processes point to a degradation mechanism involving the oxidation of the sulfur atom in the thiophene ring creating a sulfoxide side group (R-SO-R), which is then oxidized to sulfone (R-SO 2 -R) [56,61]. Signals from these groups are expected in the range of 1000-12,000 cm −1 [62][63][64]. In the case of severe overoxidation, oxidative elimination of SO 2 opens the thiophene ring leading to the formation of carbonyl groups, and even polymer chain breakage may occur with the appearance of carboxyl groups [56,61]. Since for the annealing temperature of 240 • C Raman spectra showed no evidence of carbonyl or carboxyl formation, the observed thermal degradation could be an initial stage of overoxidation, originating sulfoxide and sulfone side groups [61].
In general, the conductivity in conjugated polymers, including PEDOT, is limited by interchain charge transport, since intrachain transport along the polymer backbone is much faster. A low π-π stacking distance is required to increase interchain transport and overcome charge localization within the polymer chain [65]. The molecular π-π stacking distance depends on chain planarity and interchain packing. This is notably affected by the thermal degradation that generates side groups in the polymer backbone, and thus impairs the conductivity. In addition, side groups also reduce the chain conjugation length and increase charge localization, and, as a result, charge transport is restricted [65][66][67]. This should especially be the case for PEDOT-Al deposits annealed at 240 • C, as Raman spectra suggest more side groups at this temperature. The shift of the main peak (from 1445 cm −1 to 1451 cm −1 ) at the annealing temperature of 240 • C suggests a structural transformation of the PEDOT chains from quinoid to benzoid (Figure 5d) [68]. The quinoid conformation with its flat and straight structure favors π-π interactions between chains, whereas the benzoid structure prefers a coiled arrangement of the PEDOT chains. Therefore, at 240 • C, the π-π staking distance should increase, and the coiling of the chains would affect intrachain mobility and polymer packing. All of the above leads to limited intrachain and interchain transport, which contributes to the pronounced increase in polymer resistance [68,69].
Polymer degradation was also evidenced from the UV-Vis spectra of PEDOT:PSS-Cl and PEDOT-Al 0.50 mC. In this paper, a noticeable decrease in absorption was observed around 800 nm with increasing annealing temperature (Figure 5e). This signal has been related to electronic transitions due to the presence of polarons in the polymer. Therefore, in line with the increase in R s , the annealing temperature affects the charge carriers in the polymer [58]. Similar Raman and UV-Vis results were found for PEDOT-Al 0.750 mC ( Figure S8).
To confirm the decrease in polaronic states, a Mott-Schottky analysis was performed to determine the charge carrier density in HTMs. In conducting polymers, this is challenging as the doping level may change due to the potentials applied during the measurements. To extract information that resembles the original state of the HTMs, a low polarization electrochemical window was chosen. Figure 5f shows the results: in this plot, the slope of the linear section is inversely proportional to the charge carrier density (Table S3) [70,71]. As expected, the PEDOT:PSS-Cl and PEDOT-Al 0.50 mC 200 • C films have the highest charge carrier density. Moreover, it can be observed that the carrier density decreases as the annealing temperature increases. This reduction of charge carriers (polarons) in this polymer has been related to the formation of bonds between the chlorides (Cl − ) and positive charges on the polymer chains, and to the recombination with radicals generated during degradation [56,59,72]. Since reverse and forward sweeps showed similar results, it follows that the effect of doping was significantly reduced under these experimental conditions. Overall, Raman, UV-Vis, and Mott-Schottky characterizations reveal that the mechanism affecting the conductivity of PEDOT is accentuated at 240 • C and contributes to the noticeable increase in R s , resulting in low FF and limiting the PCE of the PEDOT-Al devices.
The effect of thermal degradation on the HTMs was also investigated through electrochemical impedance (EIS) and cyclic voltammetry (Figure 6a-c). Table S4 presents the charge transfer resistance (R ct ) values obtained by fitting the results to the equivalent circuit in Figure 6b. It was found that, for the annealing temperature of 200 • C, PEDOT-Al had almost the same R ct as pristine PEDOT:PSS-Cl, and increasing the annealing temperature also increased the R ct . In conducting polymers, the R ct is closely related to the microstructure. Thus, the symmetry and reversibility of the redox reaction depend on the overlap of the density of states (DOS) of the polymer and the redox probe, whereby a highly organized microstructure results in a DOS of the polymer that facilitates reversible charge transfer [73]. Cyclic voltammetry results for the ferricyanide redox reaction (Figure 6c) show that the peak separation (ΔEp) increases with the increase in the annealing temperature of HTMs. In turn, the redox currents decrease and the peaks are less defined. This may be attributed to a structural disorder caused by the thermal degradation of the polymer [74]. Moreover, some reports relate the increased Rct of PEDOT to its overoxidation [75] and the presence of side groups in the polymer chain [67]. Consequently, for annealing processes, a low Rct value can be taken as indicative of a high-quality conductive polymer for solar cell applications.

Impedance Spectroscopy Analysis
Finally, impedance spectroscopy (IS) was performed to compare the solar cells based on PEDOT:PSS-Cl and PEDOT-Al 0.50 mC 200 °C. In both cases, the Nyquist plots show two well-defined semicircles followed by considerable noise, which takes the form of a third semicircle as the applied potential increases ( Figure S9). Some investigations have linked the third semicircle to ionic motion in perovskite [76,77], and noise to ionic motion and reactivity of metal contacts [78]. Although the interpretation of the impedance spectra is still a matter of debate, since there is no general model for perovskite solar cells, the equivalent circuit in Figure 7a was used to fit the data [78,79]. Cyclic voltammetry results for the ferricyanide redox reaction (Figure 6c) show that the peak separation (∆Ep) increases with the increase in the annealing temperature of HTMs. In turn, the redox currents decrease and the peaks are less defined. This may be attributed to a structural disorder caused by the thermal degradation of the polymer [74]. Moreover, some reports relate the increased R ct of PEDOT to its overoxidation [75] and the presence of side groups in the polymer chain [67]. Consequently, for annealing processes, a low R ct value can be taken as indicative of a high-quality conductive polymer for solar cell applications.

Impedance Spectroscopy Analysis
Finally, impedance spectroscopy (IS) was performed to compare the solar cells based on PEDOT:PSS-Cl and PEDOT-Al 0.50 mC 200 • C. In both cases, the Nyquist plots show two well-defined semicircles followed by considerable noise, which takes the form of a third semicircle as the applied potential increases ( Figure S9). Some investigations have linked the third semicircle to ionic motion in perovskite [76,77], and noise to ionic motion and reactivity of metal contacts [78]. Although the interpretation of the impedance spectra is still a matter of debate, since there is no general model for perovskite solar cells, the equivalent circuit in Figure 7a was used to fit the data [78,79]. There is some consensus that the Rs corresponds to losses due to ohmic resistance in the cell and wiring [79]. Furthermore, in the impedance spectrum (IS), the first arc at high frequencies is attributed to the charge transfer resistance and recombination resistance (R1) [39]. The capacitance of this arc is of the geometrical type and is due to the perovskite material (C1) [80]. The second arc at intermediate frequencies is due to the recombination resistance R2, although it is usual to add R1 and R2 to estimate this resistance [39,79]. The capacitance of the second arc C2 represents the charge accumulation at the cell interfaces, which causes defects in the perovskite leading to recombination [44,80]. This capacitance increases rapidly near the maximum power point (MPP) of the solar cell. It is common to plot 1/C2 versus applied potential, as the trend resembles JV curves [44]. Due to noise, the parameters of the third arc were not taken into account. The obtained IS parameters are shown in Figure 7b-f. The Rs (9 Ω) and C1 presented similar value with both HTMs (PE-DOT:PSS-Cl, PEDOT:Al). The incorporation of Al2O3, which is a dielectric material, justifies the slight increase in capacitance (from 6.2 to 6.7 nF). It is also observed that R1+ R2 was higher for PEDOT-Al, showing that the electrochemical incorporation of the Al2O3 layer markedly increases the recombination resistance (Figure 7e).
A solar cell with a high recombination resistance will present a higher current density at an applied potential, leading to a higher Voc [31]. The recombination resistance of PE-DOT-Al decreased faster initially, but maintained a higher value compared to PE-DOT:PSS-Cl, hence the increased current density in the 0 to 0.6 V range and the higher Voc ( Figure 2a). Therefore, this suggests that the use of PEDOT-Al helps to suppress charge recombination. The increase in recombination resistance is also confirmed by the C2 capacitance values extracted from CPE2. Figure 7f shows the variation of 1/C2 with voltage for the two HTMs. A slight decrease is observed for PEDOT-Al, while for PEDOT:PSS-Cl the decrease is larger and drops abruptly at 0.5 V. A high and nearly constant 1/C2 value There is some consensus that the R s corresponds to losses due to ohmic resistance in the cell and wiring [79]. Furthermore, in the impedance spectrum (IS), the first arc at high frequencies is attributed to the charge transfer resistance and recombination resistance (R 1 ) [39]. The capacitance of this arc is of the geometrical type and is due to the perovskite material (C 1 ) [80]. The second arc at intermediate frequencies is due to the recombination resistance R 2 , although it is usual to add R 1 and R 2 to estimate this resistance [39,79]. The capacitance of the second arc C 2 represents the charge accumulation at the cell interfaces, which causes defects in the perovskite leading to recombination [44,80]. This capacitance increases rapidly near the maximum power point (MPP) of the solar cell. It is common to plot 1/C 2 versus applied potential, as the trend resembles JV curves [44]. Due to noise, the parameters of the third arc were not taken into account. The obtained IS parameters are shown in Figure 7b-f. The Rs (9 Ω) and C1 presented similar value with both HTMs (PEDOT:PSS-Cl, PEDOT:Al). The incorporation of Al 2 O 3 , which is a dielectric material, justifies the slight increase in capacitance (from 6.2 to 6.7 nF). It is also observed that R 1 + R 2 was higher for PEDOT-Al, showing that the electrochemical incorporation of the Al 2 O 3 layer markedly increases the recombination resistance (Figure 7e).
A solar cell with a high recombination resistance will present a higher current density at an applied potential, leading to a higher V oc [31]. The recombination resistance of PEDOT-Al decreased faster initially, but maintained a higher value compared to PEDOT:PSS-Cl, hence the increased current density in the 0 to 0.6 V range and the higher V oc (Figure 2a). Therefore, this suggests that the use of PEDOT-Al helps to suppress charge recombination. The increase in recombination resistance is also confirmed by the C 2 capacitance values extracted from CPE 2 . Figure 7f shows the variation of 1/C 2 with voltage for the two HTMs. A slight decrease is observed for PEDOT-Al, while for PEDOT:PSS-Cl the decrease is larger and drops abruptly at 0.5 V. A high and nearly constant 1/C 2 value represents better current density and steady-state PCE for PEDOT-Al. In contrast, PEDOT:PSS-Cl has a lower 1/C 2 value leading to a lower current density JV curve in the 0 to 0.6 V range. Impedance spectroscopy generates results more similar to steady-state results than JV curves. This is because a constant potential is applied for several minutes, whereas the JV uses a fast potential sweep. Therefore, the decrease of 1/C 2 at 0.5 V for PEDOT:PSS-Cl can be attributed to the inability of this device to maintain the current density at a constant potential above 0.5 V, thus, the lower steady-state efficiency compared to the JV PCE, since the steady-state measurements were conducted at 0.75 V [44]. It is suggested that ionic charge accumulation at the PEDOT:PSS-Cl/perovskite interface leads to a high defect density in the perovskite, whereas PEDOT-Al decreases the presence of these defects (lower C 2 values). This is supported by the Mott-Schottky analysis ( Figure S2), which showed that ions can easily penetrate the PEDOT:PSS-Cl layer causing doping/dedoping of the polymer [37]. In contrast, in PEDOT-Al, the insertion of halide ions is restricted [81], so fewer of these ions can migrate and leave the perovskite.

Conclusions
This study presented the fabrication and evaluation of inverted perovskite solar cells, using as HTM a PEDOT-Al 2 O 3 bilayer obtained by electrochemical methods. It was shown that the incorporation of electrochemical Al 2 O 3 improves the charge selectivity, passivates the perovskite, and prevents doping of the polymer. As a result, the efficiency of the solar cell increased from 12.1% to 14.3%. Notably, it was observed that the steady-state efficiency of the PEDOT-Al 2 O 3 cell is higher than the efficiency determined from JV curves, whereas without Al 2 O 3 it produces 92% of its JV efficiency in steady-state tests. The addition of electrochemical Al 2 O 3 also increased the cell stability by 2% during a 45 min continuous operation test.
The electrochemical Al 2 O 3 deposition methodology used in this work is much simpler and more affordable than the atomic layer deposition method. Nevertheless, special attention must be given to the annealing step as it was found that a high temperature treatment degrades the PEDOT (sulfur oxidation, change from quinoid to benzoid, and decrease in polarons), reducing its conductivity and cell FF. However, this effect was minimized by using an annealing temperature of 200 • C.
The results presented in this work aim to contribute to the development of advanced HTMs in perovskite solar cells based on electrochemical techniques.

Supplementary Materials:
The following are available online at https://www.mdpi.com/article/10 .3390/polym13234162/s1: Table S1: Mean and standard deviation (SD) of the photovoltaic parameters of solar cells containing different HTMs; Figure S1: Evolution of (a) PCE, (b) Jsc, (c) Voc, and (d) FF of the devices. The dots represent the mean value, and the bars show the standard deviation, 4 devices for each variation. The devices were stored under a N 2 atmosphere and exposed to 70% RH during the measurements; Figure S2 Figure S5: (a-f) AFM topographic images and RMS roughness of ITO, PEDOT:PSS-Cl, and PEDOT-Al substrates; Figure S6: (a-f) Static contact angle of water over different materials deposited over ITO; Figure S7