Perovskite-polymer composite cross-linker approach for highly-stable and efficient perovskite solar cells

Manipulation of grain boundaries in polycrystalline perovskite is an essential consideration for both the optoelectronic properties and environmental stability of solar cells as the solution-processing of perovskite films inevitably introduces many defects at grain boundaries. Though small molecule-based additives have proven to be effective defect passivating agents, their high volatility and diffusivity cannot render perovskite films robust enough against harsh environments. Here we suggest design rules for effective molecules by considering their molecular structure. From these, we introduce a strategy to form macromolecular intermediate phases using long chain polymers, which leads to the formation of a polymer-perovskite composite cross-linker. The cross-linker functions to bridge the perovskite grains, minimizing grain-to-grain electrical decoupling and yielding excellent environmental stability against moisture, light, and heat, which has not been attainable with small molecule defect passivating agents. Consequently, all photovoltaic parameters are significantly enhanced in the solar cells and the devices also show excellent stability.

Metal halide perovskite films are conventionally prepared with solution processes and subsequent crystallization. The paper describes the use of long polymeric chains, brought at early state of the processing, to increase the environmental stability and optoelectronic properties of polycrystalline perovskite solar cells. The long polymeric chains act to crosslink adjacent grains, adding to improved stability of the perovskite structure. Another impact of the polymer is to decrease the number of nucleation sites during crystallization, thereby leading to increased grain size and higher crystallinity which further improves the cell performance. The effect of long polymeric chains (polypropylene carbonate, PPC) vs (small) un-polarized molecules (ethylene carbonate, EC and propylene carbonate, PC) was investigated. These materials were chosen on the bases of their high dipole moment.
The work provides and interesting way to lock the structure and affect the grain growth by using long polymers, aiming for improved stability and performance. The analyses and arguments provided were also sound. However, the practical aspect of the finding should be addressed better.
1. The author are mainly looking into the effect of using PCC vs the effect of using EC and PC instead. A more detailed comparison to the performance of perovskites made by other groups should be given.
2. The performance (Jsc, Voc, FF, PCE) of the as-prepared cells is compared in Fig. 5. No description of the reference sample in Fig.5d was given. At least I did not find it.
3. One of the major reasons for the degradation of these perovskite solar cells is the migration of the Iodine. This has been shown in several publications, for example Journal of Physical Chemistry Letters 7, 5168 (2016), which should be cited. How the polymer can prevent the Iodine migration? The authors seem to claim that the migration of ions is reversible, but I not sure. Can the authors clarify this aspect? 4. The PCE reported in Fig.5d seem to me quite high, or too optimistic. One typical problem in measuring the PCE of such solar cells is their long relaxation times leading to a hysteresis of the J-V characteristic. (One reason is the migration of Iodine, but also other polarization mechanisms.) Many groups overestimated the PCE because their measurement was done with too short time steps and thus was influenced by transient currents. See for example Solar Energy Materials andSolar Cells 159, 197 (2017) andSolar Energy 173, 976 (2018). Can the authors show in the J-V characteristics of Fig.5f both scan directions of the voltage, forward and reverse? Is the hysteresis area reduced by using the polymer? 5. Environmental stress test was executed by placing the sample in 70% humidity for 150 h and at elevated temperature (100Â°C) in inert ambient (N2) for 66 h and exposure to light (1.5 MO) for 2 h. The performance of the cells after the environmental test was not given. It was only shown that the PCC suffered from less degradation than EC and PC (suppl. Fig. S17). Even though that does indicate that PCC will work better than EC and PC, it says little about the absolute performance of the cell. While these tests may be suitable to demonstrate a difference in using the different additives, a better comparison to a reference cell is needed.
effectively increased the ion migration activation energy from 0.40 eV (bare MAPbI3) to 0.53 eV (MAPbI3 with PPC) (Fig. R1), which can be attributed to the reduction of the perovskite's charged defects (e.g., positively charged anion vacancies) at the grain boundaries due to the passivation effect of the polymeric Lewis bases that remained in between the perovskite grains ( Fig. R2).  (Fig. R3a). When 10% of MA was replaced with FA (i.e., MA0.9FA0.1PbI3), the operational stability was improved relative to that of bare MAPbI3, and the FACs-based perovskite with a small amount of 2D perovskite (1.67% of PEA2PbI4) further increased the operational stability. The device with the FACs-based perovskite maintained ~80% of initial PCE after aging for 500 h, while MAPbI3 degraded to ~20%, which can be attributed to the better material stability against light/ high temperature and the higher tolerance for ion migration relative to that of MAPbI3. All the perovskite solar cells fabricated without PPC showed an exponential initial decay, followed by linear, stabilized decay. The devices with PPC substantially reduced the amount of this initial decrease (Fig. R4, and Table R1), which could be attributed to a reduced charged defect formation, effective defect passivation, and higher activation energy for their migration with PPC. We also extracted the T80 (time with which the PCE decays to 80% of its initial PCE) from the post-burn-in regime (linear decay regime) for all the solar cells fabricated in this work and summarized the data in Table R1. The FACs-based solar cells with PPC decreased its initial decay to 1.8%, and after 500h illumination, maintained 95.0% of initial PCE, from which the expected T80 is ~10800.1 h. We believe that this operational stability study demonstrates 1) the universality of our macromolecular intermediate phase approach for all organic cation based perovskites, and also that 2) our method can attain long-term operational stability if coupled with suitable compositional engineering.

Revised parts in the manuscript)
Page 14-15 in the manuscript, "To demonstrate the universality of our cross-linking approach, we examined the operational behavior of the devices based on various kinds of 'A'-site cations (MA, MA0.9FA0.1, and FA0.98Cs0.02). To evaluate their operational stability, all the devices were encapsulated under a nitrogen atmosphere and exposed to continuous illumination (90 ±10 mW,without UV filter) under open-circuit condition (Supplementary Fig. 26,27,and Supplementary Table 4). PPC effectively elongated the operational lifetime of all the perovskite solar cells regardless of their 'A'-site cation composition. All the devices without and with PPC showed rapid initial decay followed by slower degradation with an almost linear profile, but the devices with PPC demonstrated a less severe initial decay. Given that the initial rapid decay can be related to the migration of ion and charged defects 48 , the reduced initial decay regime can be attributed to a decreased charged defect density and an increased activation energy for ion migration as a result of the PPC-induced crystallization. Severe ion migration in perovskite solar cells can result in J-V hysteresis, and along with the possible compound formation with the electrode, could accelerate the degradation of a device during operation 49-51 . Indeed, PPC effectively increased the activation energy for ion migration in the perovskite ( Supplementary Fig. 28), resulting in reduced hysteresis than that of bare CH3NH3PbI3 ( Supplementary Fig. 29).
Furthermore, the addition of PPC slowed down the subsequent linear decay regime as well.
From this, we extracted the T80 (time with which the PCE decays to 80% of its initial value) from the post-burn-in regime (linear decay regime) for all the solar cells fabricated in this work and summarized this in Supplementary   The Nernst-Einstein relation ( ( ) = 0 ( − ), where is the Boltzmann constant and 0 is a constant) was used to calculate the activation energy for ion migration, and the lateral conduction configuration was employed for the measurement (Fig. S28 inset). Our macromolecular adduct approach, which resultantly modified the perovskite crystallization, effectively increased the ion migration activation energy from 0.40 eV (bare CH3NH3PbI3) to 0.53 eV (CH3NH3PbI3 with PPC) (Fig S28), which can be attributed to the reduction of the perovskite's charged defects (e.g., positively charged anion vacancies) at the grain boundaries due to the passivation effect of the polymeric Lewis bases that remained in between the perovskite grains, and this translated to a reduced J-V hysteresis of the solar cells with PPC.

Long chain polymer molecules have been applied to cross-link grain boudaries in
perovskite film to improve stability against moisture and oxygen in many previous reports. Nevertheless, they are missing here and even not mentioned in introduction.   P P C 0 .1 w t% P P C 0 .3 w t% P P C 1 .0 w t% P P C 3 .0 w t% P P C 5 .0 w t% r e f .  (Fig. R5).
The different influences of the polymers on crystal growth and electrical conduction were examined by analyzing the surface morphology and spatially resolved electrical conductivity of the perovskite films. As we reported in the manuscript, PPC has a strong chemical interaction with the perovskite precursors (i.e., acid-base adduct formation) (Supplementary Fig. S1-4, and  Supplementary Fig. S6, and S7). On the contrary, we found that PAA has the opposite effect on perovskite crystal growth compared to that of PPC. PAA gradually decreased the grain size as higher amounts of PAA was added (Fig. R6). All the MAPbI3 grains were finely cleaved into nanometer-sized grains upon the addition of 5.0wt% of PAA (Fig. R6c), which could be attributed to the chemical interaction between PAA and the perovskite precursors/ or solvent in the solution. Figure R6. Atomic force microscopy images of MAPbI3 film a, without, b, with 0.3wt% of PAA, and c, with 5.0wt% of PAA.
Per classical theory for homogeneous nucleation, the nucleation rate is described by using a critical free energy (∆ ), which represents the free energy required for nuclei to be stable without being dissolved in the solution and is the sum of the surface and bulk free energy of the nuclei. This critical free energy is defined as the activation energy for nucleation and used to describe the nucleation rate using an Arrhenius type equation [Chem. Rev. 114, 7610 (2014), Small 7, 2685 (2011)]: where t is time, N is number of nuclei, A is pre-exponential factor, is Boltzmann`s constant, T is temperature. The crystal free energy (∆ ) can be written as a function of surface energy , molar volume ν, supersaturation of solution S, which produce a following equation, Because the repeating functional group of PAA is a carboxylic acid (C2H3COOH) (pKa= 4.25), the PAA likely interacted with the Lewis base solvents of the precursor solution (i.e., DMSO, and DMF), instead of forming an adduct with the Lewis acidic precursors, which possibly increased the saturation level of the precursors in the solution, resulting in a decreased activation energy for crystallization (Fig. R7). As a result, the fast crystallization formed a large number of nuclei, and subsequently small crystal grains, which could severely interrupt the inter-grain electrical coupling due to the insulating nature of PAA residing in between the small perovskite polycrystals. If the PAA does not have meaningful chemical interaction with the precursors, this additive can be regarded as an inhibitor for crystal growth. As C increases, the crystallization rate exponentially decreases, thereby resulting in small grain sizes of crystals with inhibitor molecules in between the crystals.
To directly compare and visualize the electrical properties of the perovskite grains, we performed spatially-resolved electrical conductivity measurements (conductive AFM) on the perovskite films grown without/ with PC, PPC, and PAA (Fig. R8). The perovskite films were deposited on an ITO/ SnO2 electron transporting layer. To analyze their charge carrier conducting characteristics, a positive bias was applied using a Sb-doped Si tip. Both the small molecular Lewis base and polymeric Lewis base resulted in the overall electrical enhancement of the perovskite films compared to that of the bare MAPbI3 film. Particularly, the electrical conductivity in the grain boundaries was seen to significantly increase even though the film had high concentrations of the insulating organic molecules (5.0wt%). The conductivity enhancements could be attributed to an enhanced charge carrier mobility due to the higher crystallinity and an increased charge carrier concentration due to a reduction in both the defect trapping sites and non-radiative recombination in the perovskite grains. In contrast, the MAPbI3 film with 5.0wt% PAA incorporated exhibited electrically decoupled tiny grains and inhomogeneity in between the perovskite grains ( Fig. R8d). As a result, the existence of the insulating PAA in the perovskite film severely disrupted the electrical properties of the solar cells, degrading its JSC and FF significantly (Fig. R8e). Figure R8. Electrical current mapping measured by conductive atomic force microscopy of MAPbI3 a, without, and with b, 5.0wt% of PC, c, 5.0wt% of PPC, d, 5.0wt% of PAA. e, current density vs. voltage (J-V) characteristics of perovskite solar cell without, and with 5.0wt% of PC, PPC, and PAA.
Unlike small molecular additives, polymers have higher molecular length and lower molecular chain mobility, which seems to be important factors to sustain suitable interactions with the perovskite precursors and mediate the perovskite crystallization. We found that PPC, with a strong electron donating nature, strongly interacted with the perovskite precursors to form a long-range macromolecular intermediate phase. The increased activation energy for crystallization enabled enlarged crystal grains to form and the low long chain mobility of the macromolecular intermediate phase allowed the polycrystals to be crosslinked with polymerperovskite crystal composite bridges (Fig. 3). For the reason above, PPC and PAA had obviously different influences on the crystal growth and electrical properties of the perovskite films. Therefore, we believe that our inter-grain crosslinking method which involves a macromolecular polymer-perovskite composite did not lead to a severe electrical degradation of the perovskite, even when incorporated with high concentrations of the insulating additive P P C 0 .1 w t% P P C 0 .3 w t% P P C 1 .0 w t% P P C 3 .0 w t% P P C 5 .0 w t% r e f .   Per classical theory for homogeneous nucleation, the nucleation rate is described by using a critical free energy (∆ ), which represents the free energy required for nuclei to be stable without being dissolved in the solution and is the sum of the surface and bulk free energy of the nuclei. This critical free energy is defined as the activation energy for nucleation and used to describe the nucleation rate using an Arrhenius type equation [Chem. Rev. 114, 7610 (2014), T is temperature. The crystal free energy (∆ ) can be written as a function of surface energy , molar volume ν, supersaturation of solution S, which produce a following equation, Because the repeating functional group of PAA is a carboxylic acid (C2H3COOH) (pKa= 4. ation energy for crystallization (Fig. S13). As a result, the fast crystallization formed a large n umber of nuclei, and subsequently small crystal grains, which could severely interrupt the inte r-grain electrical coupling due to the insulating nature of PAA residing in between the small p erovskite polycrystals.
The reduction of crystal grain size and electrical insulation can also be considered in the aspect of inhibitor molecules by the following model [Theor. Found. Chem. Eng. 42, 179 (2008)], , f is the area occupied by the adsorbed impurity particle [m 2 ], is the growth step energy [J m -2 ], Vm is the unit volume [m 3 ] of the crystal, and is the relative solution supersaturation [dimensionless]. The equation quantifies the influence of inhibitors on the crystallization rate.
If the PAA does not have meaningful chemical interaction with the precursors, this additive can be regarded as an inhibitor for crystal growth. As C increases, the crystallization rate exponentially decreases, thereby resulting in small grain sizes of crystals with inhibitor molecules in between the crystals. The perovskite films were deposited on an ITO/ SnO2 electron transporting layer. To analyze their charge carrier conducting characteristics, a positive bias was applied using a Sb-doped Si tip. Both the small molecular Lewis base and polymeric Lewis base resulted in the overall electrical enhancement of the perovskite films compared to that of the bare CH3NH3PbI3 film.
Particularly, the electrical conductivity in the grain boundaries was seen to significantly increase even though the film had high concentrations of the insulating organic molecules (5.0wt%). The conductivity enhancements could be attributed to an enhanced charge carrier mobility due to the higher crystallinity and an increased charge carrier concentration due to a reduction in both the defect trapping sites and non-radiative recombination in the perovskite grains. In contrast, the CH3NH3PbI3 film with 5.0wt% PAA incorporated exhibited electrically decoupled tiny grains and inhomogeneity in between the perovskite grains ( Fig. S14d). As a  As the PU concentration increased (0.1wt% and 0.3wt%), the VOC and FF increased, while the JSC showed no noticeable decrease, resulting in an improved PCE (from 16.5% to 17.7%) ( Fig. R2). In contrast, the JSC and FF severely decreased (JSC: 21.3 mA/cm 2 to 17.2 mA/cm 2 , FF: 69.6% to 58.7%) as the PAA additive concentration in the perovskite precursor solution was increased (Fig. R3), which means that PAA significantly degraded the electrical properties of MAPbI3.  factor (FF) and power conversion efficiency (PCE)), b, current density versus voltage (J-V) characteristics of the best CH3NH3PbI3 solar cells with 0.1wt% and 0.3wt% PU.
We have additionally compared perovskite solar cells incorporated with three different additives (i.e., propylene carbonate (PC), polypropylene carbonate (PPC), and PAA) with different molecular structures and functional groups to examine their effect on the electrical properties of the perovskite polycrystals. (Figure R4). We explored the photovoltaic performances of the perovskite solar cells according to different additive concentrations, ranging from 0.1wt% to 5.0wt% (Fig. R4) (Fig. R4).   Fig. S6, and S7). On the contrary, we found that PAA has the opposite effect on perovskite crystal growth compared to that of PPC. PAA gradually decreased the grain size as higher amounts of PAA was added (Fig. R5). All the MAPbI3 grains were finely P P C 0 .1 w t% P P C 0 .3 w t% P P C 1 .0 w t% P P C 3 .0 w t% P P C 5 .0 w t% r e f . cleaved into nanometer-sized grains upon the addition of 5.0wt% of PAA (Fig. R5c), which could be attributed to the chemical interaction between PAA and the perovskite precursors/ or solvent in the solution. Per classical theory for homogeneous nucleation, the nucleation rate is described by using a critical free energy (∆ ), which represents the free energy required for nuclei to be stable without being dissolved in the solution and is the sum of the surface and bulk free energy of the nuclei. This critical free energy is defined as the activation energy for nucleation and used to describe the nucleation rate using an Arrhenius type equation [Chem. Rev. 114, 7610 (2014), T is temperature. The crystal free energy (∆ ) can be written as a function of surface energy , molar volume ν, supersaturation of solution S, which produce a following equation, and DMF), instead of forming an adduct with the Lewis acidic precursors, which possibly increased the saturation level of the precursors in the solution, resulting in a decreased activation energy for crystallization (Fig. R6). As a result, the fast crystallization formed a large number of nuclei, and subsequently small crystal grains, which could severely interrupt the inter-grain electrical coupling due to the insulating nature of PAA residing in between the small perovskite polycrystals.  If the PAA does not have meaningful chemical interaction with the precursors, this additive can be regarded as an inhibitor for crystal growth. As C increases, the crystallization rate exponentially decreases, thereby resulting in small grain sizes of crystals with inhibitor molecules in between the crystals.
To directly compare and visualize the electrical properties of the perovskite grains, we performed spatially-resolved electrical conductivity measurements (conductive AFM) on the perovskite films grown without/ with PC, PPC, and PAA (Fig. R7). The perovskite films were deposited on an ITO/ SnO2 electron transporting layer. To analyze their charge carrier conducting characteristics, a positive bias was applied using a Sb-doped Si tip. Both the small molecular Lewis base and polymeric Lewis base resulted in the overall electrical enhancement of the perovskite films compared to that of the bare MAPbI3 film. Particularly, the electrical conductivity in the grain boundaries was seen to significantly increase even though the film had high concentrations of the insulating organic molecules (5.0wt%). The conductivity enhancements could be attributed to an enhanced charge carrier mobility due to the higher crystallinity and an increased charge carrier concentration due to a reduction in both the defect trapping sites and non-radiative recombination in the perovskite grains. In contrast, the MAPbI3 film with 5.0wt% PAA incorporated exhibited electrically decoupled tiny grains and inhomogeneity in between the perovskite grains (Fig. R7d). As a result, the existence of the insulating PAA in the perovskite film severely disrupted the electrical properties of the solar cells, degrading its JSC and FF significantly (Fig. R7e). Figure R7. Electrical current mapping measured by conductive atomic force microscopy of MAPbI3 a, without, and with b, 5.0wt% of PC, c, 5.0wt% of PPC, d, 5.0wt% of PAA. e, current density vs. voltage (J-V) characteristics of perovskite solar cell without, and with 5.0wt% of PC, PPC, and PAA.
Unlike small molecular additives, polymers have higher molecular length and lower molecular chain mobility, which seems to be important factors to sustain suitable interactions with the perovskite precursors and mediate the perovskite crystallization. We found that PPC, with a strong electron donating nature, strongly interacted with the perovskite precursors to form a long-range macromolecular intermediate phase. The increased activation energy for crystallization enabled enlarged crystal grains to form and the low long chain mobility of the macromolecular intermediate phase allowed the polycrystals to be crosslinked with polymerperovskite crystal composite bridges (Fig. 3). For the reason above, PPC and PAA had obviously different influences on the crystal growth and electrical properties of the perovskite films. Therefore, we believe that our inter-grain crosslinking method which involves a macromolecular polymer-perovskite composite did not lead to a severe electrical degradation of the perovskite, even when incorporated with high concentrations of the insulating additive  into nanometer-sized grains upon the addition of 5.0wt% of PAA (Fig. S12c), which could be attributed to the chemical interaction between PAA and the perovskite precursors/ or solvent in the solution.  Per classical theory for homogeneous nucleation, the nucleation rate is described by using a critical free energy (∆ ), which represents the free energy required for nuclei to be stable without being dissolved in the solution and is the sum of the surface and bulk free energy of the nuclei. This critical free energy is defined as the activation energy for nucleation and used  If the PAA does not have meaningful chemical interaction with the precursors, this additive can be regarded as an inhibitor for crystal growth. As C increases, the crystallization rate exponentially decreases, thereby resulting in small grain sizes of crystals with inhibitor molecules in between the crystals. Particularly, the electrical conductivity in the grain boundaries was seen to significantly increase even though the film had high concentrations of the insulating organic molecules (5.0wt%). The conductivity enhancements could be attributed to an enhanced charge carrier mobility due to the higher crystallinity and an increased charge carrier concentration due to a reduction in both the defect trapping sites and non-radiative recombination in the perovskite grains. In contrast, the CH3NH3PbI3 film with 5.0wt% PAA incorporated exhibited electrically decoupled tiny grains and inhomogeneity in between the perovskite grains (Fig. S14d). As a result, the existence of the insulating PAA in the perovskite film severely disrupted the electrical properties of the solar cells, degrading its JSC and FF significantly (Fig. S14e)." 16.97±0.44%. As the reviewer suggested, we added the description of the photovoltaic performances of the as-prepared reference samples in the manuscript and these are also summarized in Table 2.

Revised parts in the manuscript)
Page 13 in the manuscript,  Table 2)."   (Fig. R8), which can be attributed to the reduction of the perovskite's charged defects (e.g., positively charged anion vacancies) at the grain boundaries due to the passivation effect of the polymeric Lewis bases that remained in between the perovskite grains (Fig. R9). Figure R8. Temperature-dependent conductivity of MAPbI3 film a, without and b, with PPC (inset: the schematic illustration of the lateral conduction device configuration.) Figure R9. a, Photoluminescence (PL) (Inset: normalized PL spectra) and b, time-resolved PL spectra of perovskite films without and with Lewis bases (Inset: PL lifetimes fitted from the time-resolved PL spectra).

One of the major reasons for the degradation of these perovskite solar cells is
According to a previous literature [Energy Environ. Sci. 10, 604 (2017)] cited in the manuscript, the operational behavior of devices as a function of time can be classified into two regimes: 1) initial reversible loss and 2) permanent degradation (Fig. R10a). They claimed that the reversible loss regime of perovskite solar cells is caused by the migration of ions and ionic defects because the activation energy and time scale for cation migration are much larger and longer (hours scale) than halide migration (minutes scale) (Fig. R10b-d). We included a mention on the reversible loss regime as related to halide ion migration as "Considering that the initial decay is related to the reversible migration of charged defects, the observed reduced initial decay can be attributed to a decreased defect density as a result of the PPC-induced crystal modification." As the reviewer pointed out, this sentence can mislead readers, so we revised it in our revised manuscript to avoid misunderstanding. We appreciate the reviewer's helpful comment.

[Redacted]
For the reason above, the hysteresis of the devices with PPC decreased compared to that of bare MAPbI3 (Fig. R11). The hysteresis index was calculated for standard devices with/ without PPC based on the following relation,  The reduced charged defect density and increased activation energy for ion migration with PPC were also effective in decreasing the extent of initial decay during the cell operation. We showed an exponential initial decay, followed by linear, stabilized decay. The devices with PPC substantially reduced the amount of this initial decrease (Fig. R12, and Table R1), which could be attributed to a reduced charged defect formation, effective defect passivation, and higher activation energy for their migration with PPC. We also extracted the T80 (time with which the PCE decays to 80% of its initial PCE) from the post-burn-in regime (linear decay regime) for all the solar cells fabricated in this work and summarized the data in Table R1. The FACs-based solar cells with PPC decreased its initial decay to 1.8%, and after 500h illumination, maintained 95.0% of initial PCE, from which the expected T80 is ~10800.1 h.   resulting in reduced hysteresis than that of bare CH3NH3PbI3 (Supplementary Fig. 29).
Furthermore, the addition of PPC slowed down the subsequent linear decay regime as well.
From this, we extracted the T80 (time with which the PCE decays to 80% of its initial value) from the post-burn-in regime (linear decay regime) for all the solar cells fabricated in this work and summarized this in Supplementary Table S4. All the calculated T80 lifetimes of the devices with PPC were significantly elongated. Especially, the FACs-based solar cells with PPC decreased its initial decay to 1.8%, and after 500h illumination, maintained 95.0% of initial PCE, from which the expected T80 is ~10800.1 h. The slower decay has been related to a permanent degradation of the perovskite layer accompanied by a chemical reaction and morphological change 52 . Therefore, the inter-grain cross-

Page 29-30 in the supplementary information,
The Nernst-Einstein relation ( ( ) = 0 ( − ), where is the Boltzmann constant and 0 is a constant) was used to calculate the activation energy for ion migration, and the lateral conduction configuration was employed for the measurement (Fig. S28 inset). Our macromolecular adduct approach, which resultantly modified the perovskite crystallization, effectively increased the ion migration activation energy from 0.40 eV (bare CH3NH3PbI3) to 0.53 eV (CH3NH3PbI3 with PPC) (Fig S28), which can be attributed to the reduction of the perovskite's charged defects (e.g., positively charged anion vacancies) at the grain boundaries due to the passivation effect of the polymeric Lewis bases that remained in between the perovskite grains, and this translated to a reduced J-V hysteresis of the solar cells with PPC.

Environmental stress test was executed by placing the sample in 70% humidity for 150 h and at elevated temperature (100Â°C) in inert ambient (N2) for 66 h and exposure to light (1.5 MO) for 2 h. The performance of the cells after the environmental test was not given. It
was only shown that the PCC suffered from less degradation than EC and PC (suppl. Fig.   S17). Even though that does indicate that PCC will work better than EC and PC, it says little about the absolute performance of the cell. While these tests may be suitable to demonstrate a difference in using the different additives, a better comparison to a reference cell is needed.

Answer)
We appreciate the reviewer's constructive comments. The ultraviolet-visible absorption (UV-vis-Abs) and X-ray diffraction (XRD) results in the manuscript were tested on perovskite films to investigate the influences of external degradation factors on the perovskite film with or without the effect of the polymers and their crosslinking (Fig. R13). From the results above, we confirmed that the PPC-mediated defect passivation and the enlarged/ cross-linked crystals are much more resistant to extrinsic degradation factors, namely moisture, heat, and light. To prove the better performance of the devices with PPC against these extrinsic factors, we have additionally conducted environmental stability tests on solar cell devices. [ITO/ SnO2/ Perovksite film] was exposed to moisture and high temperature to distinguish the degradation effects between the and electrode, respectively (Fig. R14).  The photovoltaic performances of bare MAPbI3 exposed to a humid environment (RH 70±5%) significantly dropped. A 50 h exposure almost fully converted the MAPbI3 to yellow colored PbI2 ( Fig. R15 inset), which can be attributed to a moisture-induced decomposition of MAPbI3  (2015)]. Most of the bare MAPbI3 devices exposed to moisture for 50 h did not work as normal photovoltaics, and the devices that worked exhibited very poor performances, with huge drops in JSC and FF (50 h exposed bare MAPbI3 device: VOC: 0.949±0.06 V, JSC: 9.85±1.34 mA/cm 2 , FF: 32.5±2.4%, and PCE: 3.08±%0.81) (Fig. R15). In contrast, the cross-linked MAPbI3 with PPC showed a much higher moisture resistance and still maintained relatively high photovoltaic performances even after 50 h exposure in a high moisture environment (50 h exposed PPC-MAPbI3 device: VOC= 1.088±0.02 V, JSC: 20.43±0.28 mA/cm 2 , FF: 72.3±1.8%, and PCE: 16.07±%0.67).  After the perovskite layers were exposed to a humid environment (RH 70±5%) (Fig. S20), the photovoltaic performances of the bare CH3NH3PbI3 solar cells dropped significantly, and the working devices exhibited very poor performances, mainly having huge drops in JSC and FF (50 h exposed bare MAPbI3 device: VOC: 0.949±0.06 V, JSC: 9.85±1.34 mA/cm 2 , FF: 32.5±2.4%, and PCE: 3.08±%0.81) (Fig. S21). In contrast, the cross-linked CH3NH3PbI3 with PPC showed much higher moisture resistance and still maintained relatively high photovoltaic performances even after 50 h exposure in the high moisture environment (50 h exposed PPC-CH3NH3PbI3 device:  PCE: 9.31±2.02%) (Fig. S22). Compared to bare CH3NH3PbI3, the PPC-CH3NH3PbI3 crosslinked devices did not show any significant drop even after heated at 150 ºC heating (VOC: 1.064±0.07 V, JSC: 21.34±0.28 mA/cm 2 , FF: 74.6±2.3%, PCE: 16.93±0.77%), and proved their better thermal stress tolerance as seen in the XRD and UV-vis-Abs results.   Figure R3. a, Photoluminescence (PL) (Inset: normalized PL spectra) and b, time-resolved PL spectra of perovskite films without and with Lewis bases (Inset: PL lifetimes fitted from the timeresolved PL spectra).

Revised parts in the manuscript)
Page 30-31 in the supplementary information, The Nernst-Einstein relation ( ( ) = 0 ( − ), where is the Boltzmann constant and 0 is a constant) was used to calculate the activation energy for ion migration, and the lateral conduction configuration was employed for the measurement (Fig. S28 inset). Our macromolecular adduct approach, which resultantly modified the perovskite crystallization, effectively increased the ion migration activation energy from 0.40 eV (bare CH3NH3PbI3) to 0.53 eV (CH3NH3PbI3 with PPC) (Fig S28), which can be attributed to the reduction of the perovskite's charged defects (e.g., positively charged anion vacancies) at the grain boundaries due to the passivation effect of the polymeric Lewis bases that remained in between the perovskite grains, and this translated to a reduced J-V hysteresis of the solar cells with PPC.

Except for PPC, it is necessary to suggest other candidates (hydrophobic polymers) for
perovskite-polymer composite approach.
Answer) As the reviewer suggested, we additionally compared the performances of perovskite solar cells incorporated with additives with different functional repeating groups, which can suggest new guidelines related to the functional group requirements of the polymers. We used three additional kinds of commercially available polymers, polyacrylic acid (PAA), poly(4vinylpyridine) (PVP), and polyurethane (PU) (Fig. R4a). All the polymers have lone pairs of electrons on nitrogen or oxygen along the polymer backbone, but different basicity and molecular dipole moments. Acrylic acid, the repeating unit of PAA, is the simplest unsaturated carboxylic As the PU concentration increased (0.1wt% and 0.3wt%), the VOC and FF increased, while the JSC showed no noticeable decrease, resulting in an improved PCE (from 16.5% to 17.7%) (Fig.   R5). In contrast, the JSC and FF severely decreased (JSC: 21.3 mA/cm 2 to 17.2 mA/cm 2 , FF: 69.6% to 58.7%) as the PAA additive concentration in the perovskite precursor solution was increased  Therefore, we conducted additional experiments on the moisture and thermal stability of the perovskite solar cells. As seen in Figure 5, we confirmed that the PPC-mediated defect passivation and the enlarged/ cross-linked crystals are much more resistant to extrinsic degradation factors, namely moisture, heat, and light. To prove the better performance of the devices with PPC against these extrinsic factors, we have additionally conducted environmental stability tests on solar cell devices.
[ITO/ SnO2/ Perovksite film] was exposed to moisture and high temperature to distinguish the degradation effects between the perovskite and the hole transporting layer (HTL) (p-doped spiro-MeOTAD) (Fig. R7). Figure R7. Schematic illustration of moisture and thermal environmental stability test of perovksite solar cells.
The photovoltaic performances of bare MAPbI3 exposed to a humid environment (RH 70±5%) significantly dropped. A 50 h exposure almost fully converted the MAPbI3 to yellow colored PbI2 ( Fig. R8 inset), which can be attributed to a moisture-induced decomposition of MAPbI3 according to the following reaction CH3NH3PbI3→CH3NH2 + HI + PbI2 [J. Mater. Chem. A 3, 8970 (2015)]. Most of the bare MAPbI3 devices exposed to moisture for 50 h did not work as normal photovoltaics, and the devices that worked exhibited very poor performances, with huge drops in JSC and FF (50 h exposed bare MAPbI3 device: VOC: 0.949±0.06 V, JSC: 9.85±1.34 mA/cm 2 , FF: 32.5±2.4%, and PCE: 3.08±%0.81) (Fig. R8). In contrast, the cross-linked MAPbI3