Solution‐Processed Metal Ion Polyelectrolytes as Hole Transport Materials for Efficient Inverted Perovskite Solar Cells

Despite achieving high efficiencies over a short time, further streamlining of hybrid lead‐halide perovskite solar cell (PSC) designs is necessary for their commercial viability. In this contribution, a new class of interfacial hole transporting layer (HTL) materials consisting of anionic polyelectrolytes comprising polystyrene sulfonate (PSS) with metal cations are explored. These materials represent alternatives to metal oxides, combining characteristics of metal oxides with the facile preparation and desirable film‐forming characteristics of polyelectrolytes. Polyelectrolytes with cations including Li, Mg, V, Mn, Co, Ni, Cu, Zn, Pd, Ag, In, Cs, and Pb as HTLs in inverted PSCs are explored. A range of positive and negative effects is observed for different metal cations, which are attributed to differences in the physical properties of the polyelectrolytes, and their influence on the electronic band structure of devices and the crystal qualities of the perovskite absorber. Ni and Cu polyelectrolytes created p‐type contacts at the anode of PSCs, improving device performance. These materials are believed to have potential in other types of devices as well. This type of metal:PSS polyelectrolyte has not yet been widely investigated, however, it is shown that it constitutes a simple and economic strategy to engineer energy band structures in perovskite devices.


Introduction
Hybrid organic-inorganic perovskite materials are a class of solution-processable semiconductors that are unique in their ability to retain excellent electronic transport properties despite significant ion migration, re-organization, and disorder. They have been shown to yield highly efficient photovoltaic and lightemitting devices from remarkably low-cost materials. [1][2][3][4] Their DOI: 10.1002/admi.202300043 high radiative efficiency, long carrier diffusion lengths, strong optical absorption, tolerance to defects, bandgap tunability, and simplicity of processing all make them attractive in a variety of optoelectronic applications including light-emitting diodes (LEDs) lasers and photodetectors. Metal halide perovskites are chromophores that exhibit not only strong light absorption but efficiently generate and transport mobile charges under illumination, a remarkable property for such low-cost materials, which gives them tremendous potential as an economical class of solar cell that can be manufactured economically and on large scales. [5] Concerted worldwide research efforts have led to frequent improvements in cell design and a rapid increase in the power conversion efficiencies (PCE) of single-junction solar cells to 25.2% over a short period of time. [6][7][8] Furthermore, perovskites are also already being implemented with great success as subcells in tandem devices with commercially established silicon technology. [9,10] In perovskite solar cells (PSCs), HTLs play a significant role in determining the collection efficiency of charges generated in the perovskite layer, and properly designed HTLs are able to decrease charge recombination at the hole-collecting electrode. [11] Three major classes of HTLs have emerged in the context of PSCs which lead to efficient devices; these include inorganic metal oxides/halides, conjugated organic

Structural Design of Target Materials
PEDOT:PSS is a highly effective HTL that is used in a variety of organic and hybrid solar cells; it exhibits outstanding performance as an electron blocking and hole transporting (with a hole mobility of 1.45 cm 2 V −1 s −1 ), characteristics which have been reported in numerous studies. [30] Although the PSS backbone is not electronically conjugated itself (benzene rings are aromatic, however, this -conjugation does not extend between adjacent repeat units), it greatly improves electronic conduction when coupled with the conjugated polymer PEDOT by compen-sating p-type charge carriers in addition to increasing the WF of PEDOT:PSS formulations. In the absence of the conjugated PE-DOT moiety, the PSS backbone alone can be considered an electrical insulator due to its wide bandgap and lack of delocalized -electrons; conduction in this kind of polymer largely depends upon the presence of ionic functionalities in its structure, which can manifest as mobile ions as well as redox-active cations which are able to mediate charge transfer through changes in oxidation state. To investigate the properties of non-conjugated PSS polymers, we have coupled a range of positively charged metal cations to the negatively charged PSS backbone to create a variety of solution-processable interlayers that, like metal oxides, are influenced by the electronic structure of the metal cations as well as the ionic character of the PSS backbone. For example, within this system, some metal cations may accept electrons from the semiconductor phase, (which may be promoted by electrochemical reaction at the anode or photoinduced electron transfer) resulting in a decrease in the oxidation state of the metal while, leaving an excess negative charge on the PSS backbone. The excess negative charges on PSS are able to compensate excess p-type carriers in the semiconductor; this makes p-type doping possible at the interface with the adjacent semiconducting layer.
There are many potential benefits of this type of material, including that it may prevent or reduce the corrosion of ITO relative to the commonly used, and highly acidic (pH 1.5-2) PEDOT:PSS, which is often a point of failure in conventional devices. The ease of synthesis and purification of PSS-based polyelectrolytes makes them affordable for large-scale production. The synthesis of metal: PSS materials simply involves the acid-base neutralization of polystyrene sulfonic acid with a metal base, such as a metal acetate, resulting in the general structures shown in Scheme 1, depending on the charge of the metal cation. These materials can then be used as HTLs or as interfacial layers, by themselves or mixed with other HTL materials, in order to change the WFs of substrates and control the Fermi energy at the active layer interface by dipole formation. Several of these structures led to dramatic increases in PSC parameters like PCE, fill factor (FF), and V OC while showing excellent electron-blocking characteristics, making them potentially useful interfacial layers in PSC and other organic semiconducting devices.
In Scheme 1 (a), M represents a metal cation and X represents an anion from a corresponding metal ionic salts (for example acetate, carbonate, chloride, acetylacetonate, etc.). These metal salts interact with polystyrene sulfonic acid where an equivalent amount of H + in polystyrene sulfonic acid is replaced by metal cations to prepare various metal:PSS polyelectrolytes. The products were synthesized according to these reactions and separated from their byproducts by precipitation based on their solubility. The pH of metal:PSS products typically increased from being highly acidic (in the case of polystyrene sulfonic acid, H:PSS) to nearly neutral due to the replacement of protons by the metal cations in the PSS structure. Among the metal cations, we have focused on the elements of groups IA, IIA, IB-VIIB, and VIII. Elements in the 7 th period were excluded from this study due to their radioactivity, while some semi-metals like Indium and Lead in group IIIA and IVA of the periodic table were also investigated. Proton nuclear magnetic resonance ( 1 H NMR) spectroscopy was employed to understand the structure of these materials. NMR studies largely showed the presence Scheme 1. Polyelectrolyte synthesis. a) General reaction used for the synthesis of metal:PSS materials and b) summary of metals used in this study. The topology of polyelectrolytes incorporating metal ions having oxidation state of c) +1 is represented by M + , d) +2 is represented by M 2+ , and e) +3 is represented by M 3+ .
of PSS backbone with peak shifts which can be attributed to the presence of metal cations. More detailed information about 1HNMR is provided in Figure S12, Supporting Information. Films fabricated from metals:PSS materials were transparent and did not possess strong visible absorption bands due to the presence of metal cations, but exhibited absorption bands in the ultra-violet region arising from benzene rings in the PSS backbone, which caused absorption features around 260 nm in all cases. The bandgap of these materials was found to be over 4 eV.
All of the metal:PSS interfacial materials were found to be much less acidic than PEDOT:PSS. The pH of PEDOT:PSS was measured to be 1.4, whereas the metal:PSS materials exhibited pH values in the range of 4.83 to 6.68 (provided in Table S1, Supporting Information) at the same concentration. We measured the stability of devices with different HTLs, using the process of long-term aging in the dark. Long-term stability data are included in Figure S1, Supporting Information. This data shows that among the different formulations, PEDOT:PSS has significantly greater stability than the other formulations. Although PEDOT:PSS only retained 35% of its initial PCE value under the same conditions, this was greater than any of the metal:PSS derivatives. Despite being pH neutral, metals:PSS interlayers alone showed noticeably decreased stability compared to PEDOT:PSS. This data demonstrates that simply neutralizing the pH of an HTL alone does not improve its stability. Cations are known to diffuse appreciably in perovskite materials. [31] In the case of metal:PSS materials, metal cations may be exchanged with Pb 2+ ions in the perovskite layer. Although the diffusion of Pb 2+ is much slower than other ions in perovskites, [31] we suspect that the instability of devices using metal:PSS polyelectrolyte HTLs as they age over several weeks can be attributed to the slow exchange of metal ions in metal:PSS materials with Pb 2+ in the perovskite film. Additionally, the hygroscopic nature of the polyelectrolytes may accelerate device deterioration. [32] In order to develop a more complete understanding of the performance of optoelectronic devices, characterizing the electronic structure at interfaces can provide valuable information. X-ray and ultraviolet photoelectron spectroscopy (XPS and UPS, respectively) are widely used to determine the chemical states of metals and electronic structures. XPS probes core energy levels and can give insight into chemical reactions that occur (by monitoring changes in oxidation or bonding state of different elements) during film processing, as well as the band bending caused by equilibration of the different Fermi levels (E F ) on both sides of the interface. [33,34] UPS is used to determine the WF of metals and the vacuum level (VL) and HOMO positions of organic semiconductors. When an organic solid contacts a metal, the organic layer may be affected by surface dipoles. Measured VL shifts can yield the magnitude and direction of the interfacial dipole (Δ). Generally, a downward shift of VL (Δ < 0) gives rise to a larger hole-injection barrier ( h ), while an upward shift (Δ > 0) provides a larger electron-injection barrier ( e ). [35,36] We investigated the electronic structures of polyelectrolytes containing metal ions. Polyelectrolytes were deposited by spin coating from a suitable solvent at an appropriate spin speed (details are provided in the Experimental Section). From the UPS spectra, the molecular orbital alignment, and band bending at the interface were determined according to the following equations; [35][36][37] where hv is the incident photon energy, E SE is the secondary edge position, E HOMO is the onset energy of the HOMO level, E g is the optical band gap, and E LUMO is the energy of the LUMO level. Energy level diagrams corresponding to the metal:PSS materials used in this study are shown in Figure 1, with red values representing HOMO (bottom) energy levels measured by UPS and blue values representing the lowest unoccupied molecular orbital (LUMO) (top) values and black values represent their WFs. The E HOMO was determined by subtracting ionization potential (IP), hv, and E SE giving values in the range of ≈7.62-8.16 eV. E LUMO was estimated by subtracting the known E g value from E HOMO , giving values in the range of ≈3.62-4.16 eV. More detailed information obtained from XPS and UPS is provided in Figures S2  and S3, Supporting Information. Analysis of these values reveals that there aren't considerable differences between the energy levels of most of the polyelectrolytes, with the exception of Cu:PSS, which possess lower LUMO and HOMO values. The bandgaps in these materials are high compared to other layers in PSCs, which leads to electrically insulating properties, however, during the fabrication of inverted-type PSCs, films with thicknesses on the order of ≈5 nm were deposited, which are thin enough to avoid increasing the series resistance of devices while still impacting the electronic band structure.
The morphology of HTLs plays a vital role in the overall operation of devices. Poor HTL morphology, such as pinholes or discontinuous films, can be detrimental to device performance since direct contact of perovskites with electrodes can short-circuit devices. Therefore, AFM was used to characterize the morphologies of all the HTLs. Figure 2 shows AFM topographic images of all the polyelectrolyte HTLs, as well as ITO and PEDOT:PSS reference samples. The morphology of ITO is somewhat rough with a relatively high root mean square (RMS) roughness value of 2.37 nm. PE-DOT:PSS has a very smooth morphology with the lowest RMS roughness value among all of the films at 0.99 nm. We observed that, although some of the morphological features looked similar to ITO, the roughness of the substrates decreased as thin films of different HTLs layers were deposited on ITO. The flake-like morphological features characteristic of ITO cover a height range of 16 nm, therefore, when HTLs with thicknesses less than 16 nm are deposited, some of the underlying morphological features of ITO may still be visible as height variations in the overlying HTL. All of the metal:PSS films exhibited RMS roughness values in the range of 0.99-1.82 nm; a few of them, such as Co, In, and Zn, had smooth morphologies with few pin holes and less evident flakelike features.
Although morphology can influence device characteristics, it is not possible to attribute all of the observed variations in device performance to morphology alone. To develop a clearer picture of how each type of polyelectrolyte impacts electrical behavior, we also employed c-AFM to investigate spatial variations in film conductivity for each HTL. These details are provided in Figures S4 and S5, Supporting Information. c-AFM is a powerful technique capable of mapping local current distribution with great sensitivity; variations in current as low as nanoamperes can be detected. Thus, the magnitude of current observed in c-AFM images is an indication of the local hole conductivity of the HTLs. In this work, PEDOT:PSS was found to exhibit the best performance among all other HTLs, however, c-AFM results showed that PEDOT:PSS films exhibited the lowest average current (0.21 nA) compared to the average currents of metal:PSS HTLs (which were in the range of 2.2 to 36.9 nA). This may be because the PEDOT:PSS films were much thicker (25 nm thick) than the metal:PSS films (0.48 to 3.04 nm thick). According to the literature, [38] it has been observed that the vertical electrical conductivity in spin-cast PE-DOT:PSS thin films is up to 3 orders lower than that along the parallel direction, due to the lamellar orientation of PEDOT and the insulating PSS [36,39] this may also contribute to the low observed conductivity in PEDOT:PSS.
Compared to the average current observed for bare ITO substrates (37.3 nA) the current was found to decrease when different metal:PSS films were spin-cast onto ITO, as expected. The average current (36.9 nA), and therefore conductivity of Cu:PSS was found to be the highest amongst all HTLs including PEDOT:PSS HTLs, in agreement with the observed device characteristics; in particular, the high FF observed in Cu:PSS devices is consistent with the high conductivity and low series resistance of this HTL.
Since the morphology of perovskite films has a significant impact on device performance, [35,36,40] we performed scanning electron microscopy (SEM) measurements to examine the effects of different metal:PSS HTLs on the morphology of overlying perovskite films. Figure S6, Supporting Information, shows the morphology of CH 3 NH 3 PbI 3 films deposited on different HTLs. It can be seen that perovskite films exhibit a non-compact morphology frequently containing pinholes on some substrates such as control devices on PEDOT:PSS, and a few HTLs like Pd:PSS and Mg:PSS. Pinholes, cracks, and surface defects tend to increase the concentration of electronic defects such as dangling bonds and vacancies. Furthermore, these pinholes can serve as non-radiative recombination centers causing significant trapassistant recombination. [39,41,42] Moreover, the pinholes are likely to give rise to shunts in solar cells due to direct contact between the cathode and anode interfacial materials, producing detrimental leakage current and low shunt resistance. The morphological study of perovskite films on different HTLs shows no major difference in morphologies, however, polyelectrolytes that do not completely cover the substrate surface showed a corresponding decrease in their device performance.
To investigate the influence of HTLs on the crystal structure of perovskite films, X-ray diffraction (XRD) patterns of perovskite films on different HTLs were collected. As shown in Figure S7, Supporting Information, there were no considerable changes in the tetragonal structure of MAPbI 3 as all the films show the same features with prominent peaks corresponding to the (110) plane. XRD parameters are summarized in Table S2, Supporting Information.
The aim of introducing different HTLs in devices is to understand their influence on the characteristics and performance of PSCs. Therefore, we examined the influence metal:PSS HTLs including, Li:PSS, Mg:PSS, V:PSS, Mn:PSS, Co:PSS, Ni:PSS, Cu:PSS, Zn:PSS, Pd:PSS, Ag:PSS, In:PSS, Cs:PSS, and Pb:PSS on device characteristics. The inverted device architecture ITO/HTL/CH 3 NH 3 PbI 3 /PC 61 BM/Al was used throughout this study. When HTLs are used as the bottom layer in PSCs, next to the transparent electrode, then the transparency of the HTLs is very important, since light must pass through the HTL before reaching the active layer. Hence, more excitons can be generated if the HTL has higher light transmittance, improving the overall performance of devices. Figure S8, Supporting Information, shows the transmittance spectra for all of the HTLs. Compared to PEDOT:PSS as a reference; all of the metal:PSS HTLs showed greater transparency throughout the ultraviolet and visible range, which indicates that more light is able to enter the active layer and hence these HTLs may increase the rate of exciton generation. Although the transmittance of the metal:PSS films was similar, all of the metal:PSS HTLs resulted in different solar cell characteristics, indicating that the identity of the metal cation had an important impact on the device performance, regardless of the transmittance through the films. Figure 3a compares normalized data corresponding to FF, short circuit current density (J SC ), and V OC for devices using different metal:PSS HTLs on ITO substrates. PEDOT:PSS and H:PSS were used as a benchmark and metal-free reference, respectively, for comparison. Table 1 summarizes the device parameters along with film thickness for each HTL. From this data, it is evident that some metal ions are not good choices as HTLs in PSCs. Pd:PSS and H:PSS, in particular, showed markedly negative effects on device characteristics, whereas Cu:PSS and Ni:PSS showed good performance approaching that of the PE-DOT:PSS benchmark. Notably, the oxides of Cu and Ni have been demonstrated to be effective HTLs and we see that polyelectrolytes based on these metal ions also perform relatively well. The most prominent difference that we observed between PE-DOT:PSS and metal:PSS HTLs was that the metal PSS HTLs tended to result in higher V OC values than PEDOT:PSS, with the exception of Pd:PSS. The V OC value obtained using PEDOT:PSS was found to be 0.89 V, while the V OC s for all of the metal:PSS devices (except Pd:PSS, V OC = 0.61 V) were found to be close to 1.0 V; the highest V OC of 1.03 V was obtained using Cu:PSS.
Among the various metal:PSS materials, the FF is the parameter that shows the greatest variation between different metal ions. Low FF is commonly attributed to poor film morphology and high h , however, given the similarities in morphology and that the h values were found to be similar for all HTLs, it appears that the ions influence the FF in additional ways as well. Among the different HTLs H:PSS showed the smallest FF of only 30%, while all metal:PSS HTLs showed higher FF values than this. PE-DOT:PSS produced the highest FF value (74%), indicating the lowest rate of recombination and fastest carrier extraction. However, Cu:PSS, Cs:PSS, and Ni:PSS all produced FF values over 50%. Because very thin films are used, electrons can back-diffuse and tunnel through the metal:PSS interlayers, which is one reason that the FF values of the materials are somewhat low compared to PEDOT:PSS. [41,43] Figure 3b summarizes the normalized PCE data for all of the metal: PSS-based devices. Here, the benchmark device (PE-DOT:PSS) showed the highest performance compared to other HTLs. Although the metal:PSS HTLs showed lower PCE compared to PEDOT:PSS, it is important to note that PEDOT:PSS is a highly optimized HTL that has been demonstrated to offer excellent performance in tens of thousands of studies on PSC, organic solar cells as well as LEDs , and higher performing HTLs are not often encountered. Although the metal:PSS HTLs do not yield higher PCE than PEDOT:PSS, they provide a valuable platform to compare the effect of metal ions on device performance. In our series of new HTLs, the champion HTL material was found to be Cu:PSS, which yielded the best performance among all other metal:PSS, including a relatively high V OC of 1.03 V, a FF of 59% and PCE of 12.8% which is comparable to the PCE of the PE-DOT:PSS device at 14.9%. The ability of Cu and other metal ions to undergo reversible oxidation and reduction allows the HTLs to transport charges, helping them to improve the hole extraction and consequently the J SC in PSCs when used as HTLs. However, despite the redox activity of the metal ions, the polymer backbone (PSS) is relatively insulating, leading to relatively low overall conductivity in the polyelectrolytes.
Films of the metal:PSS HTLs were deposited at roomtemperature and annealed at variable temperatures to identify the influence of annealing on device parameters and identify optimal annealing conditions. Current density-voltage (J-V) curves and external quantum efficiency (EQE) spectra were obtained and shown in Figures S9 and S10, Supporting Information.
Since charge carrier recombination has an immense impact on the observed J SC and FF of PSC devices, J-V curves were collected under variable light intensity to gain insights into carrier recombination. The FF of solar cells is determined by how efficiently charge carriers are collected at electrodes prior to recombination. Since different types of recombination exist, it is informative to distinguish between different recombination mechanisms in order to understand the reasons for low observed FF values in PSCs and identify strategies to improve them. Several recombination processes are well-known in PSCs, including radiative recombination, non-radiative recombination, and interface recombination. [44] Furthermore, nonradiative recombination includes Auger recombination, trap-assisted recombination (Shockley-Read-Hall recombination), electron-phonon interactions, and carrier-carrier scattering. [44] The dissociation of excitons depends strongly on the strength of the electric field within the device and consequently influences geminate recombination rates, where photogenerated excitons fail to completely dissociate into free charge carriers. The influence of geminate recombination on the FF is well-known, however, recently, nongeminate recombination processes such as Shockley-Read-Hall type (also known as trap-assisted recombination), bimolecular re-combination, and surface-trap assisted recombination have been identified as performance-limiting loss mechanisms and have been receiving increasing attention in the context of organic solar cells and PSCs. These processes can be probed with simple variable light-intensity experiments. Such data are summarized in Figure 4 which shows the light intensity and J-V characteristics for all the HTLs used in this study under variable illumination intensity. In this figure, we compare representative metal:PSS polyelectrolytes with metal cations having oxidation states in the range of +1 to +3, however, a complete set of metal:PSS recombination data are provided in Figure S11, Supporting Information. Additionally, ITO without any HTL and PEDOT:PSS are included as references for comparison. Cs:PSS, Zn:PSS, and In:PSS have oxidation states in the range of +1, +2, and +3, respectively (as shown in Scheme 1). Figure 4a shows the J SC as a function of light intensity on a logarithm scale, which should follow a power-law dependence according to the formula J SC ∝ (P light ) . [45] If the value of is less than 1, it indicates the presence of bimolecular recombination. [45] The data in Figure 4a shows that the J SC values follow almost a linear relation with light intensity, and the values for all HTLs are close to 1, suggesting that few carriers are lost by bimolecular recombination for any of the HTLs. The H:PSS, Co:PSS, and In:PSS and Pd:PSS HTLs show values of ≈ 0.9, Figure S11a,b, Supporting Information, indicates that these HTLs may induce slight losses due to bimolecular recombination compared to the other HTLs. However, because for these devices is 0.9, this does not appear to be a dominant loss mechanism.
Similarly, the light intensity dependence of the V OC can be measured to determine which types of recombination are dominant. The value of slope obtained from a linear fit of the V OC versus the natural logarithm of light intensity may indicate which type of recombination is dominant; bimolecular recombination is implied if the value of the slope ≈ K B T/q, (where, K B is Boltzmann constant, T the temperature and q the elementary charge); trap-assisted recombination is implied if the slope is greater than K B T/q and surface trap-assisted recombination is dominant in the case that the slope is less than K B T/q.
The value of the slope was found to be 1.2 K B T/q in the case of ITO and Pd:PSS; the value of slope in Mn:PSS, Ni:PSS, In:PSS, and Cs:PSS was found to be 1.7 K B T/q, while slopes of 1.8 K B T/q and 1.6 K B T/q were observed for Cu:PSS and Ag:PSS, respectively, as shown in Figure S11c Figure 4d shows the contribution of geminate recombination to overall FF losses. The contribution of both geminate and nongeminate recombination can be used to quantify losses in FF. For an ideal solar cell, the maximum obtainable FF is estimated to be 90% for MAPbI 3 , according to the Shockley-Queisser limit. At low light intensity, the recombination is predominantly geminate since there are few excitons (resulting in a low population of electrons or holes) available for other recombination pathways. Therefore, the reduction in FF at low light intensity is attributed to geminate recombination. [46] Geminate recombination is more severe in all HTLs as shown in Figure S11e, Supporting Information, compared to PEDOT:PSS. This indicates that the electric field necessary to disassociate excitons into mobile carriers is strongest in PEDOT:PSS compared to all of the metals:PSS HTLs. The metal:PSS HTLs are all relatively thin compared to PEDOT:PSS and contain smaller total numbers of ionic charges, which therefore results in a weaker electric field across the perovskite active layer than in PEDOT:PSS and results in lower values of J SC and FF compared to PEDOT:PSS. FF losses from non-geminate recombination are shown in Figure 4c. Here, the difference in the value of FF at 1 and at 100 mWcm −2 corresponds to non-geminate recombination losses. As in Figure 4d it can clearly be seen that in PEDOT:PSS the difference is just 8%, which is the lowest among all the HTLs used in this study. Further details are provided in Figure S11f, Supporting Information. Hence, it indicates that similar to geminate recombination, nongeminate recombination is the lowest in PEDOT:PSS among all HTLs. Overall, Non-geminate recombination in all other metals:PSS is consistent with the observed device performance.
Despite being capable of high PCE values, PSCs are known for being relatively unstable compared to other types of solar cells, and thus, their stability is a primary concern for practical applications. PEDOT:PSS contains a large proportion of unreacted sulfonic acid groups, and this acidity is thought to contribute to the instability of PSCs when it is used as an HTL. [44] We evaluated the stability of metal:PSS HTLs and compared them to the acidic H:PSS control film as shown in Figure S1, Supporting Information. The stability of devices with each HTL was measured by taking J-V curves intermittently over 75 h and the reported parameters are taken as the average of these. We observed that devices using H:PSS degraded the most quickly, as expected, however, metal:PSS HTLs degraded more quickly than PEDOT:PSS, despite having more neutral pH values. We attribute this degradation to the diffusion of metal ions from the ultra-thin metal:PSS layers into the ionic perovskite layer over time. However, a detailed understanding of degradation mechanisms in the HTLs and interfaces under electric fields and with thermal energy is necessary to develop an accurate picture of how metal ions interact with the perovskite layer over time, which lies outside of the scope of the present work. We anticipate that by improving the conductivity of the polyelectrolytes it will be possible to use thicker layers and mitigate the problems associated with low FF fill factors and temporal stability.

Conclusion
In conclusion, we have demonstrated a simple, cost-effective, highly transparent, easily synthesized, series of solutionprocessable HTLs that can be used in inverted PSCs. The strategy used to design the material was to target interfacial materials that can support p-doping in an adjacent semiconducting layer by incorporating redox active metal cations with an anionic polyelectrolyte, allowing the material to remove electrons from the intrinsic semiconductor, leaving an excess of p-type carriers in the semiconductor compensated by an excess of negative charges in the polyelectrolyte at the interface. Furthermore, by keeping all aspects of device fabrication identical, and using a common polyelectrolytic anion (PSS), this study provides unique insight into how different metal cations affect the behavior of HTLs and the energy band structure of solar cell devices. Among the metal:PSS polyelectrolytes, Cu:PSS performed best due to its p-type nature as revealed by high WF, and high conductivity (shown in c-AFM) despite having similar morphologies and properties compared to other metals:PSS layers. Ni:PSS was also found to function relatively well as an HTL. These observations are consistent with the use of Ni and Cu oxides as HTLs in other studies. Another important point we observed is that Pd:PSS had a consistent and dramatically negative effect on device parameters. Considering that many types of organic HTLs used in PSCs are prepared by Pd-catalyzed cross-coupling reactions, this observation highlights the importance of carefully removing any traces of Pd from organic HTLs. The free acid H:PSS also showed consistently negative effects on device characteristics and the lowest stability. Although the benchmark PEDOT:PSS performed better than the metal:PSS HTLs alone, commercial PEDOT:PSS does contain a significant proportion of acidic H:PSS and this result implies that it may be possible to improve the performance and stability of PEDOT:PSS by removing or replacing the excess H:PSS used in PEDOT:PSS formulations.
A brief summary of some of the key beneficial findings in this study is as follows: 1) Metal:PSS HTLs are a useful platform to investigate the influence of PSS on electronic band structure without PEDOT. We see that PSS polyelectrolytes generally create a p-type junction in adjacent semiconductor layers even without PEDOT. 2) We are able to compare the effect of different metal ions on HTL behavior without changing any other part of the device.
Notably, metal ions that are known to form effective p-type oxide HTLs (especially Cu and Ni) produce the most effective p-type polyelectrolyte interlayers as well. In contrast, Pd 2+ and H + consistently and dramatically degraded devices. 3) All the metal: PSS interfacial materials were found to be much less acidic than PEDOT: PSS. 4) All the metal: PSS HTLs possess very high transparency (generally 99-100%T throughout the visible spectrum) compared to PEDOT: PSS which indicates that more light can enter the active layer and hence these HTLs have the potential to increase exciton generation in the active layer. 5) The most prominent difference that we observed in device parameters between PEDOT: PSS and metal: PSS HTLs was that the metal PSS HTLs tended to result in higher V OC values than PEDOT: PSS, (with the notable exception of Pd: PSS). 6) A great benefit of metal: PSS interlayers is that they also exhibit good performance without thermal annealing or other processing.
This study of metals:PSS materials explored a fair number of metal cations, however, it only explored one type of polyelectrolyte (PSS), and investigated only ultra-thin films with one metal ion at a time; this type of metal polyelectrolyte represents a new class of interlayer with significant possibilities for the exploration of more diverse combinations of ions and polymer backbones.
Metal Synthesis: Metal:PSS polyelectrolytes were synthesized using the following processes.
Li:PSS: Li 2 CO 3 was dissolved in 2 mL deionized water in a 10 mL vial. HPSS was slowly added to Li 2 CO 3 solution which bubbled vigorously producing CO 2 gas, consistent with Equation (4). After the bubbles stopped evolving, the mixture was shaken vigorously several times and allowed to stand overnight. The solution was filtered through a 0.45 μm cellulose acetate syringe filter and precipitated in isopropyl alcohol. 0.5 mL of hexane was added to help break the milky colloid. The mixture was then centrifuged, and the supernatant liquid was discarded. The polymer was re-dissolved in 10 mL of methanol and filtered through a 0.45 μm PTFE filter to remove any unreacted Li 2 CO 3 . The clear solution was precipitated into ethyl acetate, and hexane was added to help break the milky colloid. The precipitated polymer was washed with isopropyl alcohol and hexane, then it was dried in a nitrogen glovebox. The pH of a solution (2.1 mg/ 5 mL H 2 O) was 7.2.
Mg:PSS: A solution of MgCl 2 was made by dissolving 1.45 g MgCl 2 in 8 mL methanol. 5 mL of this solution was added to 3 mL of HPSS solution to form a clear solution (Equation (5)). This solution was diluted with 15 mL of isopropyl alcohol which caused the polymer to precipitate. The precipitated polymer was centrifuged first and then re-dissolved in 1.5 mL of aqueous MgCl 2 solution and precipitated by diluting with 20 mL methanol. The process of dissolution and precipitation was repeated again with the remaining 1.5 mL MgCl 2 . The generation of HCl from HPSS was an equilibrium process, thus, the repeated interaction with MgCl 2 was intended to shift the equilibrium in favor of Mg:PSS production by repeatedly removing any HCl that was produced and adding excess Mg ions. After the third precipitation, the polymer was re-dissolved in 5 mL H 2 O, and reprecipitated in isopropyl alcohol twice, followed by washing with isopropyl alcohol and hexane, then drying in a nitrogen glovebox. The pH of a solution (4.8 mg/ 5 mL H 2 O) was 5.9, indicating close to 100% replacement of H + ions with Mg 2+ ions.
V:PSS: Vanadyl acetylacetonate (VOacac) 568 mg was dissolved in 15 mL methanol and 3 mL of HPSS was added to the solution. (Equation (6)) The mixture was stirred for 1 h at room temperature. The product was precipitated by adding diethyl ether which resulted in a green dispersion. The dispersion was then centrifuged, and dark green color polyelectrolyte was isolated from the bottom of the centrifuge tube. It was soluble in water, methanol and slightly soluble in IPA. More diethyl ether was added to wash the product several times. The dark polymer was separated and allowed to dry under vacuum. A dark green color powder was obtained after vacuum evaporation of residual solvents overnight. pH of the solution (5 mg/ 5 mL H 2 O) was 5.08.
2HPSS + VO(acac) 2 → VO(PSS) 2 (ppt) +2acac (6) Mn:PSS: Mn acetate 846 mg was dissolved in 15 mL of water. Afterwards, 3 mL of HPSS was added to the solution and the dispersion broke down resulting in a homogeneous mixture (Equation (7)). The mixture was sonicated for 20 min. After that, the mixture was precipitated using isopropyl alcohol. It was subjected to centrifuging to separate the precipitated polyelectrolyte. After centrifugation, the material had a viscous gel-like consistency and was transferred to a different vial. The viscous material was kneaded under IPA with a spatula to extract the remaining water until a fibrous polymer was formed. The solid polyelectrolyte was separated from the liquid and dissolved in water. It was then reprecipitated using IPA. The solid was then washed with additional IPA and dried under vacuum. The pH of a solution (5 mg/ 5 mL H 2 O) was 6.03.
2HPSS + Mn(CH 3 COO) 2 → Mn(PSS) 2 (ppt) +2CH 3 COOH (7) Co:PSS: 118.94 mg CoCO 3 was taken into a beaker and 2 mL of 2 m concentrated HCl was added dropwise. CO 2 gas was evolved after the acidbase reaction and the color changed from pink to dark blue forming the [Co(H 2 O) 6 ]Cl 2 complex (Equation (8)). 5 mL deionized water was added to the reaction media which dissolved the blue Co-complex and unreacted CoCO 3 was removed by filtration. 2 mL of HPSS was added to the filtrate and it was ultrasonicated for 20 min (Equation (9)). IPA was added to precipitate the polyelectrolyte which was separated by centrifuging. The product was washed with IPA and dried under vacuum overnight. The pH of a solution (5 mg/ 5 mL H 2 O) was 6.02.
Ni:PSS: Ni-acetate (0.729 g) was dissolved in 25 mL of a DIwater/methanol (1:1) mixture. 3 mL of HPSS solution was added and the mixture was precipitated in isopropyl alcohol/ethyl acetate/hexane and centrifuged (a thick aqueous phase separated) (Equation (10)). The dense, green gel was re-dissolved in water, then precipitated in IPA/ethyl acetate/hexane again. The viscous gel was kneaded with a spatula under IPA to remove excess water from the gel and yield a solid material, then washed with IPA and dried under vacuum. The pH of a solution (5 mg/5 mL water) was 5.4.
Cu:PSS: Cu acetate (0.309 g) was dissolved in 25 mL of DIwater/methanol (1:1) mixture. 3 mL of HPSS solution was added and the mixture was precipitated in isopropyl alcohol, to which ethyl acetate (5 mL) and several drops of hexane were added (to induce liquid / liquid phase separation) and the mixture was centrifuged (a viscous aqueous phase separated) (Equation (11)). The dense, blue-green gel was redissolved DIwater followed by precipitation in isopropyl alcohol/ethyl acetate/hexane again. The viscous gel was kneaded with a spatula under isopropyl alcohol to yield a solid material. Then the separated solid was washed with isopropyl alcohol and hexane. The pH of a solution (4.8 mg/ 5 mL H 2 O) was 5.90.
Zn:PSS: Zn-acetate (0.643 g) was first dissolved in 30 mL of deionized water and then 3 mL of HPSS was added to the solution, causing a dispersion to change to a homogeneous mixture. The mixture was sonicated for 20 min (Equation (12)). After that, isopropyl alcohol/ethyl acetate/hexane (15/5/5 mL) was added to the mixture in order to precipitate the polyelectrolyte. The solution turned whitish, and it was centrifuged to separate the precipitated polyelectrolyte. A viscous gel was recovered after centrifuging. The viscous material was kneaded under IPA to extract excess water until a fibrous polymer was formed. Then it was dissolved in water and re-precipitated in IPA. The solid was then washed with additional IPA and subjected to drying under vacuum. The pH of a solution (5 mg/ 5 mL H 2 O) was 6.45.
Pd:PSS: Palladium acetate (0.219 g) was dissolved in 24 mL of methanol and 1 mL of HPSS solution was added (Equation (13)). The mixture was sonicated for 20 min and then it was precipitated in IPA and centrifuged. A dark brown precipitate was separated which was dissolved in distilled water and re-precipitated in IPA again. The solid was separated and washed with IPA several times and dried under vacuum. The solid obtained could only be dissolved in water. The pH of a solution (5 mg/ 5 mL H 2 O) was 5.04.
Ag:PSS: 489 mg of Ag acetate was dissolved in 200 mL of water in a 500 mL flask at 80°C using an oil bath. 3 mL of HPSS solution was added to the Ag acetate solution after removing the flask from the oil bath (Equation (14)). The mixture was precipitated in a mixture of isopropyl alcohol and hexane (10:1 volume ratio) and centrifuged. The dark-brown gel was kneaded under isopropyl alcohol to yield a solid material. After separating the solid, it was washed with warm MeOH several times. The solid was dried under vacuum overnight to yield a dry product. The pH of an aqueous solution (5 mg/ 5 mL H 2 O) was 5.51.
HPSS + AgCH3COO → AgPSS + CH 3 COOH (14) In:PSS: A solution of InCl 3 was made by dissolving 221.18 mg InCl 3 in 8 mL deionized water. 5 mL of this solution was added to 3 mL of HPSS solution to form a clear solution (Equation (15)). This solution was diluted with 15 mL of isopropyl alcohol which caused the polymer to precipitate. The precipitated polymer was centrifuged first and then re-dissolved in 1.5 mL of InCl 3 solution and precipitated by diluting with 20 mL deionized water. The process of dissolution and precipitation was repeated again with the remaining 1.5 mL InCl 3 solution. After the third precipitation, the polymer was re-dissolved in 5 mL H 2 O, and re-precipitated in isopropyl alcohol several times, followed by washing with isopropyl alcohol and hexane, then drying under vacuum overnight. The pH of an aqueous solution (5 mg/ 5 mL H 2 O) was 4.83.
3HPSS + InCl 3 → In(PSS) 3 + 3HCl (15) Cs:PSS: Cs 2 CO 3 was dissolved in water in a 50 mL vial. HPSS was slowly added to Cs 2 CO 3 which bubbled vigorously resembling CO 2 gas evolution as a reaction byproduct (Equation (16)). After the bubbles stopped, the mixture was shaken vigorously several times and allowed to stand overnight. The solution was filtered through a 0.45 um cellulose acetate syringe filter and precipitated into IPA. A little bit of hexane was added to help break the milky colloid which formed. The mixture was centrifuged, and the supernatant liquid was discarded. The polymer was re-dissolved in 10 mL of water and filtered through a 0.45 um PTFE filter again to remove any unreacted Cs 2 CO 3 . The clear solution was precipitated into IPA, and hexane was added to help break the milky colloid, however, only a small amount of polymer was separated upon centrifuging the milky colloid. The precipitated polymer was washed with IPA and hexane, then dried under vacuum overnight. The pH of an aqueous solution (5 mg/ 5 mL H 2 O) was 6.89.
2HPSS + Pb(CH 3 COO) 2 → Pb(PSS) 2 (ppt) +2CH 3 COOH Device Fabrication: Inverted planer structure PSCs were fabricated with the structure of ITO/HTL/MAPbI 3 /Al. ITO-coated glass substrates were cleaned with detergent solution and then sequentially cleaned ultrasonically with deionized water, acetone, and isopropyl alcohol for 20 min each. The ITO substrates were dried using a hot air gun and then treated in UV ozone for 10 min before use. Different HTLs were employed for making devices. A variety of HTLs including H:PSS, Li:PSS, Mg:PSS, V:PSS, Mn:PSS, Co:PSS, Ni:PSS, Cu:PSS, Zn:PSS, Pd:PSS, Ag:PSS, In:PSS, Cs:PSS, and Pb:PSS were next deposited. All of the HTLs were deposited from aqueous solutions with concentrations of 0.01 wt.%. All the HTLs were spin-cast at 2000 rpm in air, and subsequently annealed at 100°C. MAPbI 3 films were prepared using a previously reported procedure. [46] Briefly, perovskite films were deposited via a solvent engineering method by spin coating a precursor solution in two steps at 3500 rpm for 30 s and then at 6500 rpm for 5 s. Anhydrous chlorobenzene (60 μL) was dripped at the center of the substrate during the second step. The prepared films were then placed on a hot plate at 90°C for 10 min under an atmosphere of N 2 . After depositing the perovskite layers, PC 61 BM was spin-coated at 2000 rpm for 30 s and annealed at 60°C for 10 min. To complete the devices, 100 nm of Al was deposited by thermal evaporation under a vacuum of 1 × 10 −6 Torr.
Device Characterization: The J-V characteristics of the solar cell devices were measured using a Keithley 2635 source measure unit under A.M 1.5G illumination with an irradiation intensity of 100 mW cm −2 . Simulated solar light intensity obtained from a Xe arc lamp was calibrated using a certified reference silicon photodiode immediately before testing. EQE measurements were conducted using a QEX7 system manufactured by PV Measurements, Inc. Device stability was measured for unencapsulated devices in an air atmosphere with 40% to 45% relative humidity with intermittent exposure to simulated 100 mW cm −2 solar illumination during the measurement of device characteristics.
Other Characterization: UPS spectra were obtained using a Thermo Fisher Scientific ESCALAB 250XI. UV-vis absorption spectra were collected using an Agilent Carry 500 UV-vis spectrometer. AFM images were obtained using an INOVA Multimode microscope operating in tapping mode.

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