Modulation of Ionically Generated Space Charge Effects at Hybrid Perovskite and Oxide Interfaces via Surface Modification

Interfacial space charges significantly influence transport and recombination of charge carriers in optoelectronic devices. Due to the mixed ionic‐electronic conducting properties of halide perovskites, not only electronic effects, but also ionic interactions at their interfaces need to be considered in the analysis of space charges. Understanding of these interactions and their control is currently missing. This study elucidates the ionic effects on space charge formation at the interface between methylammonium lead iodide (MAPI) and alumina, and its modulation through surface modification using organic molecules. Embedding insulating alumina nanoparticles within MAPI films leads to enhancement of the electronic conductivity. This effect is consistent with the formation of an interfacial inversion layer in MAPI and can only be explained on the basis of ionic interactions. Such an effect is attenuated by surface modification of the oxide via the chemisorption of organic molecules. Finally, the same trend is observed in solar cells, where reducing the potential of the distributed space charges within the composite active layer improves device performance. These findings emphasize the necessity of taking into account ionic interactions to control the space charge formation at interfaces involving mixed ionic‐electronic conductors, an essential aspect in the performance optimization of halide perovskite‐based devices.


Introduction
Mixed ionic-electronic conduction is a fundamental and crucial property of halide perovskites, with important implications for DOI: 10.1002/admi.202300874their application as solar cell materials.In these compounds, significant ion migration occurs even at ambient temperature and is dominated by the transport of halide vacancies with possible minor contributions from other ionic defects. []t is widely recognized that the migration of ions within the halide perovskite layer plays a crucial role in the electrical response and performance of solar cells, as evident from hysteresis and phase segregation phenomena observed in these devices. [2]Consequently, understanding and harnessing ion transport in halide perovskites is an essential prerequisite to enhance the performance, reliability, and stability of perovskite solar cells (PSCs).In this context, a remaining question is the role of mixed conduction in the interfacial space charge equilibrium at halide perovskite junctions with contacting materials, an important factor in the electronic charge carrier dynamics within the device, including charge extraction and recombination. [3]nterfacial optimization of perovskite solar cells is one of the most critical aspects that can lead to significant performance improvement in the future development of this device class.Surface modification strategies aimed at reducing trap concentration [4] (passivation) and improving energy level alignment [5] have been extensively explored in the past years.The effect of controlling interfacial band bending via doping has also been addressed. [6]In all these instances, the conventional picture describing interfacial equilibration involving electronic charge carrier equilibrium, through Fermi level alignment, is the commonly used framework to analyze mixed conducting halide perovskite devices. [7]n the other hand, ionic equilibrium at interfaces has been largely overlooked.A previous study by Kim et al. demonstrated  the presence of a significant ionically generated space charge at the interface between MAPI and alumina. [8]Because alumina is electronically insulating, these results imply that ionic effects can have a major role in space charge formation at junctions involving MAPI.This finding emphasizes the importance of considering ionic effects in the interpretation and prediction of space charge equilibrium in halide perovskite-based devices, as already established in other systems based on ionic or mixed ionic-electronic conductors. [9]In particular, since mobile ionic defects are majority carriers in MAPI, the determining role of ionic interactions in the space charge equilibrium may be of more general validity, raising the question of how such interactions could be appropriately tuned. [10]urface modification techniques, such as sensitization with self-assembled monolayers (SAMs), are widely used in the field of halide PSCs in order to tune interfacial properties. [11]These techniques are commonly aimed at reducing charge recombination and improving charge extraction from the perovskite layer to the charge transporting layer. [12]In some cases, possible effects of the SAM on the ionic properties of devices have been discussed.For example, G. Tumen-Ulzii et al. conducted a study wherein they explored the application of a fullerene derivative self-assembled monolayer. [13]The researchers proposed that the presence of the SAM reduces the localization of positive ions from halide perovskite to SnO 2 , consequently leading to a remarkable reduction in hysteresis within the devices.Despite their widespread use in PSCs, the role of organic sensitizers in the formation of interfacial space charges in PSCs is largely overlooked, particularly with regard to ionic effects.
In this study, we investigate how systematic variation of ionic interactions at halide perovskite interfaces can be achieved, and we quantify the resulting changes in the interfacial space charge potential.We focus on the interface between MAPI, used here as a representative halide perovskite compound, and alumina.Because of the insulating properties of alumina, any observed changes in the space charge potential in MAPI can be attributed solely to ionic interactions.By comparing the MAPI/alumina interface with and without modification of the oxide surface using organic molecules, we are able to test the impact of the SAM on the ionic behavior and subsequent formation of the interfacial space charge (see hypothesis in Figure 1).In this experiment, we adopt organic molecules widely used for interface engineering, [2-(3,6-Dimethoxy-9H-carbazol-9-yl)ethyl]phosphonic acid (MeO-2PACz) and benzylphosphonic acid (BPA).After establishing the formation of a stable monolayer on the alumina surface, we investigate the changes in the surface chemistry of alumina on surface modification and the properties of composite films where nanoparticles of the oxide are embedded in MAPI.We employ electrochemical techniques to determine the sign and magnitude of the space charge potential formed in MAPI when it comes into contact with alumina, both with and without surface modification using organic molecules.Finally, we fabricate devices where the presence of controlled ionically generated space charges results in distributed junctions with n-type regions forming at the MAPI/alumina interfaces in the active layer.We investigate the solar cell performance as a function of the built-in potential due to these space charges and discuss the implications in terms of device design.

Oxide Surface Modification and Modulation of Ionic Interactions
The preparation of surface-modified alumina particles and the fabrication of composite films are described in the Experimental section.Since MAPI films are fabricated via solution processing, it is necessary to check whether the molecules can be effectively anchored onto the alumina surface and withstand exposure to the relevant solvent used in the perovskite film preparation (Dimethyl sulfoxide, DMSO).To test this, Fourier-transform infrared spectroscopy (FTIR) measurements are carried out on the alumina particles coated with the organic molecule (Figure 2a,b) after washing with Tetrahydrofuran (THF, used for the surface modification) and after an additional DMSO washing step.The FTIR spectrum of alumina with MeO-2PACz surface modification exhibits distinct peaks when compared with the baseline referring to the bare alumina.These include a peak at 1576 cm −1 , which corresponds to the asymmetric stretching of the ring with the methoxy group.Additionally, peaks within the range of 1493-1436 cm −1 indicate the aromatic C═C in-plane stretch, in the 1329-1289 cm −1 range the stretching vibration of the carbazole ring, 1246 cm −1 C─N stretching in the carbazole group, 1130-1174 cm −1 and 1040-1059 cm −1 P═O and P─OH functional groups.The BPA surface-modified alumina revealed the presence of peaks at 1496 and 1452 cm −1 , indicating the presence of aromatic C═C bonds, and in the 1040-1250 cm −1 range associated with P─OH and P═O functional groups.These tests confirmed that FTIR peaks from the organic molecules were still detectable even after DMSO washing, indicating that the molecules remained stably anchored to the alumina surface.
In order to examine the effect of MeO-2PACz and BPA on the ionic interaction at the surface of Al 2 O 3 in MAPI precursor solution, zeta potential measurements are performed in a DMSO solution containing MAPI related salts (Figure S1, Supporting Information).The value of the zeta potential of alumina (2 wt.% particle dispersion in solution) in 0.5 m Pb(NO 3 ) 2 /DMSO solution is 40 mV, consistent with a tendency for Pb cations to be adsorbed on the oxide surface. [8]The zeta potential decreases to ≈12-15 mV for the surface-modified Al 2 O 3 , indicating that the presence of MeO-2PACz and BPA molecules reduces the adsorption of Pb cations onto the alumina surface in this solution.Such reduced ionic adsorption could be caused either by the steric effects due to the organic molecules adsorbed on the alumina surface or from alterations in the surface basicity of alumina caused by the phosphonic acid groups. [14]-ray photoelectron spectroscopy (XPS) measurements of a MAPI and composite films (nominal oxide volume fraction 0.8 vol.%) on SiO 2 substrates were conducted to gain further insights into the properties of surface-modified alumina and possible changes in the ionic interaction between MAPI and either bare or surface-modified alumina (Figure 2c,d).In the case of the Al 2 O 3 :MAPI composite film (referred to as A:MAPI below), an Al peak was detected, whereas no such peak was observed for the MeO2PACz-alumina:MAPI (MeO:MAPI) or the BPA-alumina:MAPI (BPA:MAPI) films.Also, a peak shift appeared in the A:MAPI film for the signal associated with oxygen.Such shift is however not present in the samples with surface-modified alumina.While coverage of the alumina particles surface with organic molecules is expected to reduce the Al signal, its complete disappearance is most probably due to a difference in surface morphology, with more of the oxide particles being present at the surface for the A:MAPI film compared to the MeO:MAPI and BPA:MAPI films.Tendency for aggregation of alumina particles in the A:MAPI sample is also corroborated by the SEM images (see below and Figure S4, Supporting Information).We can rule out differences in oxygen exposure for the samples, as all films were fabricated and measured together.We therefore conclude that the O 1s peak in the MAPI, MeO:MAPI, and BPA:MAPI films is attributed solely to oxygen contaminations (e.g., hydroxides, water) during sample preparation.Because the oxygen signal for the pure alumina particles is shifted to lower binding energies, the O 1s peak for the A:MAPI sample is likely to be due to a combination of such contaminants and the alumina particles.As for the C, N, Pb, and I peaks, only slight changes in magnitude with no significant changes in peak position were observed across the different samples (Figure S2, Supporting Information).This is expected given the relatively small concentration of the oxide particles, making changes in peak intensities due to interfacial effects difficult to detect.Inductively coupled plasma optical emission spectroscopy (ICP-OES) measurements confirmed the presence of the particles in these films, indicating their actual volume fractions to be 1.6 vol.% for both the A:MAPI and MeO:MAPI, and 1.8 vol.% for BPA: MAPI.
The crystal structure of MAPI and composite films deposited on alumina substrates with Au electrodes was analyzed using Xray diffraction (XRD) (Figure S3, Supporting Information).All films displayed peaks that correspond to the tetragonal structure of MAPI, without significant PbI 2 peaks, confirming the absence of significant residue from the fabrication process or of degraded MAPI products.No peaks associated with oxide particles were detected either, possibly due to the dispersion of small particles (less than 50 nm) with a low volume fraction in films.
The film morphology was assessed using a scanning electron microscope (SEM) (Figure S4, Supporting Information).All films, including the MAPI control, showed similarly sized crystals ranging from 180 to 200 nm.The A:MAPI film showed enhanced roughness compared with the control MAPI sample.This might be due to the tendency of the alumina particle to aggregate within the film, giving rise to a slightly less homogeneous morphology.Interestingly, the composite films MeO:MAPI and BPA:MAPI exhibited reduced roughness in comparison to A:MAPI film, which could be attributed to improved dispersion of the coated particle in the solution and the film compared to the bare particles.Crosssectional SEM images revealed that MeO:MAPI and BPA:MAPI films are also slightly thicker than the A:MAPI and MAPI films.

Measurements of Electronic and Ionic Conductivities of Composite Films
MAPI thin films as well as A:MAPI composite films with or without surface modification are fabricated on alumina substrates with interdigitated gold electrodes in order to measure electronic and ionic conductivities.1a,1b,8,15] Figure 3 shows the resulting partial conductivities plotted for the case of argon (very low p(I 2 )) and the fixed p(I 2 ) case.Previous experimental results investigating the film thickness dependence of such conductivity measurements for MAPI films deposited on similar horizontal device structures show a dominant contribution of the bulk to the measured conductivity. [8]Therefore, the values reported in Figure 3 and the rest of this work can be attributed to the bulk properties of the samples.
In MAPI, the electronic conductivity ( eon ) increases when changing the atmosphere from argon to the lowest fixed value of the iodine partial pressure by a factor of about  p(I 2 ) ∕  Ar = 60.On the other hand, the ionic conductivity ( ion ) is almost constant regardless of the atmosphere.] Schottky or anti-Frenkel defect pairs (involving methylammonium vacancies or iodide interstitials as compensating defects, respectively) have been previously discussed as a possible dominant ionic disorder, [17] while an extrinsic situation is also possible [8] and will be used in the analysis below.
For the A:MAPI film, a qualitatively similar behavior is observed, however: i) a shallower slope is detected for the  eon dependence on p(I 2 ) compared to MAPI and ii) a low value of  p(I 2 ) ∕ Ar is recorded, due to a strikingly large value of  eon under argon.When analyzing the data for the composite films containing surface-modified alumina particles, both observations i) and ii) are less evident, with these films behaving in a largely similar fashion as the MAPI control sample.
All films exhibited good reversibility when the iodine partial pressure was lowered again.However, after the reversibility test for A:MAPI, the conductivity value was lower than the starting value obtained under argon.This observation suggests that there may have been a slight degradation of the MAPI film, although no clear peak corresponding to lead iodide in the XRD pattern was observed after the conductivity measurement (Figure S5, Supporting Information), indicating that the crystal phase remained mostly stable.Nevertheless, we cannot exclude possible minor degradation occurring in the film.
In Figure 4, we show the value of  p(I 2 ) ∕ Ar for the electronic and ionic conductivity of all samples.Such a parameter referring to the electronic conductivity shows a significant drop from >60 in MAPI to <2 when including alumina particles in MAPI, and its recovery to considerable values (≈20-40) when the alumina surface is covered with the organic molecules.Importantly, the ionic conductivity change when measuring each sample in argon or under fixed p(I 2 ) is in the order of 1-2 and approximately unvaried across samples.
In order to interpret the results obtained in Figures 3 and 4, we refer to the defect chemical model describing the bulk properties of MAPI and, in addition, consider the interfacial contribution to the conductivity.We refer to the  eon trend plotted in Figure 3 and extrapolate an estimated value for the iodine partial pressure of the argon condition relevant to our experiment.By fitting a line to the data obtained for MAPI under fixed iodine partial pressure between 17 nbar to 2 μbar with the (theoretical, see below) slope of 0.5 on a double-logarithmic plot, we obtain a value of 6 pbar for the p(I 2 ) at which the measured conductivity of MAPI under argon would be expected.This is essentially assuming that, even under argon, holes are the majority of electronic charge carriers in the bulk of MAPI.Based on this, the observations described above become even more evident (see Figure 5; Figure S8, Supporting Information, for ionic conductivity results).The A:MAPI composite film shows a shallow slope in the iodine partial pressure range from 17 nbar to 2 μbar and, at very low p(I 2 ) (argon), it shows a large electronic conductivity (in order of 10 −8 S cm −1 ), which cannot be explained based on the expected dependence of the hole concentration in the bulk on p(I 2 ).This result can be explained by considering the contribution of an interfacial layer with enhanced conductivity.In particular, because such enhancement is present only at very low p(I 2 ), it must be due to electrons that are accumulated in the proximity of the interface with alumina, an effect that disappears when increasing the p(I 2 ).To emphasize this point, the expected dependence of the electron conductivity at the interface versus p(I 2 ) is shown in Figure 5, based on the measured data point in argon and with a slope of −0.5.The observed "inversion" behavior, whereby MAPI is p-type in the bulk but becomes n-type when approaching the interface with alumina, points toward a significant space charge potential drop in MAPI close to the interface.Since alumina is an insulator, this effect (heterogeneous doping) must be caused by ionic interaction (equilibration) between MAPI and alumina.
Figure 5c,d shows that the "inversion" effect disappears when the surface of alumina is covered by MeO-2PACz or BPA.Indeed, the electronic conductivity measured at very low p(I 2 ) for the MeO:MAPI and BPA:MAPI can be well described by the contribution of the holes in the bulk, as for the case of MAPI.While inversion is not reached at the lowest p(I 2 ) in these films, the overall lower value of the conductivity in these samples compared with MAPI may still indicate the presence of an interfacial space charge potential.Because such potential changes would be smaller in the films with surface-modified alumina than in the one with bare alumina, it would only cause depletion of holes in the regions close to the particles (see below) reducing the overall conductivity.The activation energy (E a ) measured for both ionic and electronic conductivities supports this interpretation.
In Figure 6, we show measurements performed during an increasing temperature followed by a decreasing temperature scan.All films showed slightly lower electronic conductivity than the original value in the final reversibility test at low temperatures.This might be caused by slight degradation of MAPI after measurements performed at high temperatures, even though no significant degradation was detected in the XRD obtained after the experiment (Figure S5, Supporting Information).Here we comment, on the main trends in terms of the fitted E a , which hold within the reported uncertainty.In A:MAPI, the E a of electronic charge carriers is lower compared to MAPI for the first temperature scan, consistent with the presence of an "inversion" space charge at the interface between MAPI and alumina, although such a difference becomes negligible during the reverse scan.The E a of ionic carriers in A:MAPI is comparable to the one in MAPI, indicating that bulk conduction of ionic defects is possible in the composite sample, possibly due to the aggregation and isolation of alumina particles (Figure S5, Supporting Information, see also comparable values of ionic conductivity for MAPI and A:MAPI and impedance analysis in Figure S7, Supporting Information).MAPI composite films with MeO-2PACz and BPA covered alumina show higher E a of electronic charge carriers compared to the A:MAPI case, indicating that no "inversion" space charge effect is present in these samples.Finally, the larger E a of the ionic charge carrier in these samples compared with the A:MAPI film may be a signature for the "deple-tion" space charge zone at the interface between surface-modified alumina and MAPI.
So far, the results indicate that the presence of the MeO-2PACz or BPA on the surface of alumina reduces the interfacial ionically generated space charge in MAPI due to a reduced ionic interaction between the perovskite and the alumina particles, as also supported by the analysis of zeta potential results described earlier.Differences in the dielectric environment due to the presence of the molecules would not be sufficient to explain the observed trend.The space charge required to compensate for a fixed charge adsorbed on the alumina surface would be independent of the presence of a molecular interlayer.We, therefore, hypothesize that steric blocking of ionic species in MAPI by the molecular modifiers is responsible for the observed change in space charge properties.We note that we cannot exclude a change in the point of zero charge of the surface on adsorption of the phosphonic acid groups, which could also contribute to such observation.

Interpretation of Conductivity Measurements in Terms of Space Charge Formation
Based on the conductivity measurements presented above, the space charge potentials expected at the different interfaces are calculated using a simplified model.Figure 7 illustrates a qualitative scheme showing the charge carrier concentration profile as a function of the distance from the interface with the alumina particles.The diagrams emphasize the influence of the space charge potential on charge carrier redistribution at such interface for the case of alumina:MAPI composite, as well as the case of surfacemodified alumina with MeO-2PACz and BPA.The schematics in Figure 7a show that the bulk of MAPI is intrinsic (the concentration of electrons and holes is much lower than the concentration of iodide vacancies).Figure 7b highlights the formation of a space charge region of width * forming in proximity to the MAPI alumina interface.In our previous work, we inferred a value of * ≈ 20 nm, by assuming a Mott-Schottky situation at such interface, whereby the depletion of positive iodide vacancies leaves behind the negative (fixed) compensating charge (concentration in the order of 10 18 cm −3 ). [8]Electrons and holes are accumulated and depleted, respectively, based on the electrostatic potential profile that is determined here by the ionic equilibration.As depicted in the figure, the presence of the molecules leads to a reduction in the variation of defect concentrations near the alumina interface compared with the bulk values.
For the A:MAPI sample, the enhanced value of  eon at low p(I 2 ) compared to the MAPI film can be used to estimate the interfacial space charge potential in the MAPI side.According to the analysis presented in reference, [8] one can write for the measured electronic conductivity ( eff ) due to the bulk conduction and the interfacial accumulated electrons where  ∞ represents the bulk electronic conductivity (dominant hole contribution),  n∞ is the conductivity in the bulk due to electrons, Ψ a denotes the volume fraction of the particles, Ω a represents the surface-to-volume ratio of the oxide particles,  represents the Debye length (see section 5 in supporting information), F is Faraday constant, R is the ideal gas constant and T is temperature.Additionally,   (assumed to be 0.5 here) is a geometrical factor describing the percolation of space charge conduction in proximity to the particles' surface.The equation yields a space charge potential of ≈840 ± 80 mV in MAPI at the interface with alumina.This value is in approximate agreement with the previous estimation. [8]n the case of MeO:MAPI, a similar analysis using the measured electronic conductivity under argon conditions reveals a reduced space charge potential of ≈610 ± 30 mV.This reduction can be attributed to the presence of organic molecules on the surface, which leads to a decrease in the adsorption of Pb at the interface between alumina and MAPI.Because BPA:MAPI shows a smaller  eon than the MAPI sample, it is not possible to determine a space charge potential using this method, possibly due to the lack of percolation pathways for bulk conduction.If therefore we use the value of hole conductivity under argon extracted based on the ideal slope of 0.5 (as indicated by the red dotted lines in Figure 5) as  ∞ , we obtain 840 ± 80, 740 ± 40, and 670 ± 20 mV for the case of A:MAPI, MeO:MAPI, and BPA:MAPI, respectively.
Another method to calculate the space charge potential in these samples involves the evaluation of the depletion situation expected for iodide vacancies and holes on their conductivity.Equation ( 2) is used in this case, [8] from which we estimate the electronic conductivity of the depletion layer ( sc ) based on the expected bulk conductivity ( ∞ corresponding to the value obtained for the MAPI sample) and the measured conductivity of the composite samples ( eff ) at high p(I 2 ) (to minimize any contribution from electron conduction).
From the calculated value of  ∥ sc , we estimate Δϕ = 120 and 160 mV for MeO:MAPI and BPA:MAPI, respectively.Note that this value is a lower limit to the space charge potential (see Section S6, Supporting Information).Because our data suggest that hole conduction in the A:MAPI sample is not blocked by space charge regions between MAPI and alumina, we cannot apply this analysis to this sample.Despite the simplified nature of the analysis, the reduction of the space charge potential forming on the MAPI side when the alumina is modified with molecular monolayers is evident from the described trend.

Implications for Solar Cell Devices
The results reported in this study suggest that ionically generated space charges forming at the interface between alumina and MAPI can lead to changes in the conduction type in the perovskite at the interface between the two phases.This concept is of potential interest for applications where the combination of a p-type and an n-type region is needed within the active layer of devices.Here, we explore whether such p-n junction, which can be obtained in a distributed fashion within composite films, is useful for solar cells.Previous reports have suggested that internal potential fluctuations, [18] as well as Rashba splitting may underlie the efficient charge separation observed in the bulk of hybrid perovskites under illumination. [19]Later reports seem to point to the relatively low exciton binding energy as a sufficient reason to explain such observation. [20]To date, limited success in controlling homogeneous doping in this class of material has made it challenging to verify the role of built-in potential within the active layer on the performance of solar cells.Here, we address such a question by resorting to heterogeneous doping, [9,21] whereby the use of insulating alumina particles embedded in the p-type MAPI film can create local n-type regions.
PSCs with MAPI, A:MAPI, MeO:MAPI, or BPA:MAPI active layer were fabricated based on the procedure described in the Experimental section.First, we ascertain that a similar effect as the one reported in the previous section is present also in these devices, where different contact layer configurations are used compared to the horizontal interdigitated electrode configuration used above.Because of the contact selectivity of the electrodes, not only transport but also interfacial and recombination resistances in the solar cell devices need to be considered in the evaluation of the conductivity measurement of the perovskites under dark and close to equilibrium.The apparent electronic and ionic conductivity in solar cell devices measured by DC Galvanostatic polarization and AC impedance measurement are shown in Figure S9 (Supporting Information).The electronic conductivity values are lower than the value extracted from the horizontal devices, consistent with a large recombination resistance, which is effectively in series with the transport resistance in the bulk of the active layer.Interestingly, the A:MAPI composite device exhibits the highest apparent electronic conductivity in this series of devices, consistent with the results obtained for the horizontal structures.This trend can be attributed to the formation of a percolating inversion space charge layer at the interface between MAPI and alumina, which provides an efficient conduction pathway for electrons at the MAPI/alumina interface between the two contacts.The electron holes are transported through the MAPI bulk in all samples and are effectively depleted in proximity to the electron transport material (ETM, here mesoporous TiO 2 ). [8]In order to check this hypothesis further, the electronic conductivities of unipolar devices using MAPI or A:MAPI composite as an active layer are compared (Figure S10, Supporting Information).The conductivity values obtained in this case confirm the presence of electron conduction in the active layer through the space charge zone between alumina and MAPI, also confirming that such space charge regions are percolating in the MAPI layer from bottom to top (from the TiO 2 interface side to the spiro-OMeTAD side) in the composite based solar cell devices.Finally, the lower electronic conductivity obtained for the composite-based devices with MeO-2PACz or BPA surface-modified alumina compared with the A:MAPI case indicates that electron conduction through the film is significantly reduced, due to the expected lower value of the interfacial space charge potential.In summary, these results point toward the opportunity of testing the effect of the space charge potential forming between MAPI and alumina in comparison with a traditional MAPI-based solar cell, while also being able to exclude effects related to composite morphology, by also testing the composite solar cells with surface modified particles.
In order to evaluate the effect of such properties on solar cell performance, current-voltage curves are measured under 1 Sun equivalent illumination (Figure S11, Supporting Information).We report the relevant photo-conversion parameters for both forward and reverse scan directions in Figure S12 (Supporting Information).The measurements exhibit significant hysteresis for all samples.As the trend across samples is independent of the scan direction, in Figure 8, we present the parameters referring to the reverse current-voltage scan only.These are plotted as a function of the space charge potential between MAPI and oxide that is estimated using Equations (1 and 2) for horizontal devices.The results indicate that the presence of the distributed and percolating p-n junction within the active layer is not beneficial to the working mechanism of the solar cell.The lower short circuit current density (J SC ) recorded for the A:MAPI device compared with the MAPI and, importantly, also with the MeO:MAPI and BPA:MAPI devices suggest that any possible improvement in charge separation due to the presence of a space charge in the active layer is not reflected in a larger collected current.Besides the J SC , the open circuit voltage (V OC ) and the fill factor in the A:MAPI devices are the lowest recorded among the cells, suggesting that the presence of a distributed p-n junction does not lead to reduced recombination or improved transport within the device.We note that we obtain a similar trend also when analyzing solar cells with lower (0.3 vol.%) particle concentration (Figure S13, Supporting Information).Once again, because all parameters seem to partially recover for surface-modified alumina particles, we are able to exclude the role of morphology in the observed trend.Time-resolved photoluminescence (trPL) measurements of MAPI, A:MAPI, MeO:MAPI, and BPA:MAPI thin films on quartz substrates are measured to evaluate the effect of recombination on the observed trend of solar cell performance (Figure S14, Supporting Information).Based on the similar recombination dynamics of the samples, such a trend cannot be explained by differences in recombination within the active layer.We suggest that the decrease in performance for the A:MAPI is due to enhanced surface recombination at the interface between MAPI and the HTM layer where, due to ionically generated space charge regions, an increased electron concentration is expected.The fact that A:MAPI shows a more pronounced presence of alumina particles on the film's surface compared with the MeO:MAPI and BPA:MAPI thin films (see discussion of XPS data above) might also contribute to such observation.
Because the ionic effects described here rely on specific surface interactions, we expect them to play a role in the determination of the space charge equilibrium at the interface between halide perovskites and other contact materials used in solar cells.Moreover, our findings indicate that the surface modification method used in this work, and widely employed in perovskite solar cells, [4][5][6]22] can also be used as a method to tailor such interfacial ionic effects. Surfce modification techniques would therefore open possibilities for designing space charge zones at interfaces with contact materials from an electronic but also an ionic point of view.The technique also offers a promising approach for manipulating the distribution of p-or n-type transport regions within the active layer of halide perovskites through heterogeneous doping.While our results show that obtaining such a distributed p-n junction reduces performance in solar cells, the same approach could pave the way to new device functionalities.

Conclusion
We provide evidence that space charges at halide perovskite interfaces can be dominated by the interaction of mobile ionic defects in the perovskite with the surface of the contact phase.We  estimate a value of ≈800 mV for the space charge potential forming in MAPI in proximity to its interface with alumina, consistent with our previous report.We demonstrate that such value can be reduced significantly (≈100-600 mV) by modifying the surface of the oxide with organic molecules that form a stable and compact monolayer.These findings highlight the influence of organic molecules on the space charge potential and conductivity.When surface modification techniques are employed in halide perovskite devices, it is crucial to consider not only how the monolayer can alter the energy level alignment between contacting materials, but also how they may selectively block or interact with specific mobile ionic defects, leading to modulation of the space charge potential at the interface.The results point to strategies to control distributed p-n junctions within composite films by taking advantage of the heterogeneous doping effects induced by ion interactions with contact surfaces.We showed that the presence of such distributed built-in potential is not beneficial to the performance of solar cells.While other uses of such architectures may be possible (e.g., field-effect or bipolar transistors, thermoelectric devices), we also expect that similar control of ionic interactions occurring between halide perovskite and semiconducting contact materials can provide additional tools in designing interfacial space charges with significant implications on device optimization.
1a,16b] Argon was used as a carrier gas and was flown in a quartz container with iodine.The container was kept in a thermostat at a fixed temperature.The iodine partial pressure was calculated based on the equilibrium pressure of iodine at the thermostat temperature. [23]urface Modification of Alumina: MeO-2PACz (16.8 mg) and BPA (8.5 mg) were added separately to 50 mL of anhydrous THF each.Then, 0.5 g of alumina nanoparticles was added to 10 mL of MeO-2PACz or BPA/THF solution and left for 24 h in the dark.The particles were separated by centrifugation at 8000 rpm for 8 min.Following centrifugation, the separated particles were mixed with 20 mL of THF and centrifuged again.The particles were then dried at 60 °C and subsequently ground in a mortar in air.Next, the dried particles were mixed with 20 mL of DMSO and centrifuged.Finally, the separated particles were dried at 110 °C with mortar grinding in air.
Perovskite Precursor Preparation-For Horizontal Devices: PbI 2 (2.075 g, 4.5 mmol) and MAI (0.715 g, 4.5 mmol) were mixed with 3 mL of DMSO.This solution was then heated at 50 °C until all the powders were dissolved.The solution was cooled down to r. t. and filtered using a 0.45 μm syringe filter.Alumina particle (7.08 mg) was mixed with the filtered 1 mL of 1.5 m MAPI solution.
Perovskite Precursor Preparation-For Solar Cells: PbI 2 (2.014 g, 4.4 mmol) was put into mixed solvents of 2.4 mL of DMF and 0.6 mL of DMSO.The solution was heated at 90 °C until all powders were dissolved.The solution was then cooled down and mixed with MAI (0.668 g, 4.2 mmol).This solution was used without a separate filtering process.For oxide:MAPI composite layer, 4.56 mg of alumina particle was mixed into 0.4 mL of solution.
Device Fabrication-Horizontal Devices: An interdigitated gold electrode pattern was fabricated on a sapphire (0001) substrate with a finger width of 5 μm, spacing between the fingers of 10 μm, and thickness of 200 nm.In order to ensure the adhesion between the gold layer and sapphire, a 20 nm thick layer of chromium was evaporated prior to the gold evaporation.After cleaning the substrate with plasma, 30 μL of the prepared perovskite precursor solution was dropped on the cleaned substrates, under an argon gas atmosphere.The following spin coating program was used: acceleration time 1 s, spin speed 65 rps, duration 1 s; acceleration time 1 s, spin speed 150 rps, duration 180 s.When 165 s were left, 500 μL of chlorobenzene was dropped.
Device Fabrication-Solar Cell Devices: Solar cell devices were fabricated at the Laboratory of Photonics and Interfaces (Prof.Micahel Grätzel) at EPFL.FTO-coated glass substrates (4 mm thick) were chemically etched using zinc powder and 4 m HCl.The substrates were then washed in deionized water.They were sonicated with Hellmanex 2% solution in deionized water.Following cleaning in water, they were sonicated with acetone and ethanol for 10 min each.Finally, the air was blown on them for drying.The solution of the compact TiO 2 layer was prepared with 600 μL of titanium diisopropoxide bis (acetylacetonate), 400 μL of acetylacetone, and 14 mL of ethanol.The substrates were placed on a hotplate with a metal mask.Then the temperature of the hot plate was increased up to 450 °C.The solution was sprayed with oxygen as a gas carrier.After spraying, the temperature of the hot plate was kept at 450 °C for 20 min, followed by slow cooling of the substrate to room temperature.The substrate was cleaned by oxygen plasma, then 50 μL of TiO 2 paste:EtOH (1:8 weight ratio) solution was dropped on the c-TiO 2 layer and spin-coated at 4000 rpm for 20 s under ambient conditions.The sample was subsequently transferred to a hot plate and heated at 80 °C for 5 min, then the sample was sintered as follows: 125 °C for 5 min, 325 °C for 5 min then 450 °C for 30 min.After this process, the temperature was slowly cooled down to room temperature, then the substrate was cleaned by oxygen plasma.The substrates were moved to a dry-air box.A 50 μL of the perovskite precursor solution was spin-coated on the substrate by followed program under dry air: acceleration 200 rpm s −1 , speed 1000 rpm for 10 s, acceleration 2000 rpm s −1 , speed 5000 rpm for 30 s.A 200 μL of chlorobenzene was dropped when 15 s was left.The film was annealed at 100 °C for 5 min.Spiro-OMeTAD solution (90 mg of spiro-OMeTAD, 23 mg of LiTFSI (520 mg LiTFSI in 1 mL of acetonitrile) in 39 μL of tBP, in 1 mL of chlorobenzene) was spin-coated on perovskite layer at 4000 rpm for 30 s and dried under dry air at r. t. under ambient light.Finally, 80 nm of Au electrode was fabricated by thermal evaporation.
Characterization: Zeta potential was measured using DT-1200 spectrometer (Dispersion Technology, Inc., Quantachrome) at room temperature.MAI, PbI 2 , or Pb(NO 3 ) 2 were dissolved in DMSO first and mixed with the oxide nanoparticles at 2 wt.%.Right after the dispersion of oxide particles, the zeta potential was measured under an ambient atmosphere.Kratos Zxis Ultra, monochromated Al K (h = 1486.6eV) was used for XPS measurement.The sample was moved from the argon glove box to XPS using an airtight sample holder without exposure to air.The ICP-OES measurements were performed using SPECTRO CIROS.The MAPI and composite films were dissolved in nitric acid.Calibration was carried out with multi-element solutions mixed with standard solutions or single-element solutions.XRD patterns were checked with an X-ray diffractometer (Empyrean Series 2, PANalytical GmbH) in reflection geometry with a Cu K radiation energy of 40 kV and a current of 40 mA.The samples were mounted in a domed airtight sample holder to keep the Ar atmosphere while measuring.SEM was operated using Zeiss Merlin with a secondary electron detector at EHT 3 kV.DC polarization was performed using a source meter (Keithley model 2634B) in the dark, at a controlled temperature and gas atmosphere.AC impedance spectroscopy was measured using Alpha-A (Novocontrol) with a frequency range from 1 MHz to 1 mHz, and an AC voltage of 0.02 V. Electronic and ionic conductivity were obtained by methods that were previously reported in reference. [15]Currentvoltage characteristics were recorded under ambient temperature and air conditions.The PSCs were measured using a 300-W Xenon light source (Sol3A) from Newport.The spectral mismatch between AM 1.5G and the solar simulator was calibrated by a Schott K113 Tempax filter (Prazosopms Glass & Optik GmbH).A silicon photodiode was used as a light-intensity calibrator for each measurement.Keithley 2400 was used for the currentvoltage scan by applying an external voltage bias and measuring the response current with a scan rate of 50 mV s −1 .TrPL was measured with a LifeSpec II fluorescence spectrometer (Edinburgh Instruments) with a picosecond pulsed diode laser (EPL-510, Edinburgh Instruments, pulse width of 85 ps).

Figure 1 .
Figure 1.a) Schematics showing the effect of blocking ionic interaction through surface modification using organic molecules on charge carrier redistribution at a MAPI-oxide interface.b) Molecular structure of BPA and MeO-2PACz, which are used in this study for the surface modification of aluminum oxide.

Figure 2 .
Figure 2. Results of FTIR measurements on alumina particles with surface modification using a) MeO-2PACz and b) BPA, after washing with THF and after a second washing step using DMSO.c,d) XPS spectra of alumina particles, MAPI, and its composite films with bare and surface-modified alumina.All composite films are fabricated with 0.8 vol.% nominal oxide particle concentration (estimated value of 1.6 vol.% in A:MAPI, MeO:MAPI, and 1.8 vol.% in BPA:MAPI by ICP measurement).

Figure 3 .
Figure 3. Electronic and ionic conductivity measurement in argon (very low p(I 2 )) and different fixed iodine partial pressure of a) MAPI, b) A:MAPI, c) MeO:MAPI, and d) BPA:MAPI composite films.The empty symbols show the reversibility of the measurements when going from high to low p(I 2 ) (H→LP).

Figure 4 .
Figure 4. Ratios of the electronic and ionic conductivity values measured under argon and fixed p(I 2 ) = 17 nbar.

Figure 5 .
Figure 5. (Above) Energy level diagram of MAPI in different iodine partial pressure and (bottom) measured electronic conductivities of a) MAPI, b) A:MAPI, c) MeO:MAPI, and d) BPA:MAPI.Red dotted lines with an ideal slope of 0.5 are fitted to the measured data under fixed iodine partial pressures in the range between 17 nbar and 2 μbar.The iodine partial pressure of the argon atmosphere is extrapolated based on this line in the MAPI case and used to visualize the data of the other samples.The blue dotted line in (b) is plotted based on the conductivity data point measured in argon with an ideal slope of the interfacial electronic contribution of −0.5.The slope for each case is written next to the fitting line using brackets.Blue solid lines are fitted to the data based on the expected p(I 2 ) dependence of the (bulk) hole and (interfacial) electron conductivities.

Figure 6 .
Figure 6.Results of electronic and ionic conductivity measurement in p(I 2 ) = 17 nbar and temperature from 30 to 60 °C for a) MAPI, b) A:MAPI, c) MeO:MAPI, and (d) BPA:MAPI.The activation energies associated with the low-to-high temperature scan (H → LT) and the high-to-low temperature scan (H→LT) are shown next to each dataset (the H→LT value is shown in brackets).

Figure 7 .
Figure 7. Sketch of mobile charge carrier concentrations in a) MAPI, b) A:MAPI, c) MeO:MAPI, and d) BPA:MAPI in the bulk and at the interface, based on the interpretation of the data in Figure 4.The parameter * indicates the width of the space charge region.

Figure 8 .
Figure 8. (Top) J SC , V OC , FF, and PCE (from reverse J-V scans shown in Figure S11, Supporting Information) of solar cell devices displayed against the estimated space charge potential at the interface between MAPI and alumina particles w or w/o surface modification (The forward and reverse scan results are shown in Figure S12, Supporting Information).The electron transport material (ETM) is TiO 2 and the hole transporting material (HTM) is Spiro-OMeTAD in this study.Detailed information about device fabrication can be found in the Experimental section.For the MAPI solar cells without alumina particles, a value of Δϕ = 0 V is used.(Bottom) A sketch of a partial solar cell structure consisting w or w/o surface modification of alumina:MAPI composite active layer.The MeO-2PACz and BPA are indicated by white dots on the surface of alumina particles.Separate P and N conduction pathways are generated due to space charge formation at the interface between MAPI and alumina.The surface modification using MeO-2PACz and BPA reduces the space charge potential between and MAPI, resulting in less pronounced N-type conduction at the interfaces.