Perovskite Solar Module: Promise and Challenges in Efficiency, Meta‐Stability, and Operational Lifetime

Perovskite photovoltaics (PVs) are an emerging solar energy generation technology that is nearing commercialization. Despite the unprecedented progress in increasing power conversion efficiency (PCE) for perovskite solar cells (PSCs), up‐scaling lab‐made cells to solar modules remains a challenge. In this work, the recent progress of making perovskite mini‐modules is reviewed. In particular, a database summarizing the module size, performance, hysteresis, and operational lifetimes reported in the literature is built. After analyzing the performance losses from scaling PSCs to mini‐modules based on the data collected from the literature, the current key to high‐performance perovskite mini‐modules is found to be the coating method optimization. If the perovskite layer quality is well reserved, a >24% mini‐module efficiency is projected by only considering the losses from lateral resistivity and laser scribing area. Next, performance characteristics are explored including hysteresis and meta‐stable power outputs that must be overcome to correctly characterize perovskite modules. Finally, current challenges associated with the long‐term stability of perovskite modules are examined and the importance of such durability for commercialization is discussed. It is hoped that the findings in this review provide a bridge for the development of perovskite modules that will lead to commercialization in the near future.


Introduction
To reduce carbon emissions, solar energy is one of the most promising renewable energy sources capable of supplying the DOI: 10.1002/aelm.202300093world's rising demand for energy. [1]Despite an 85% reduction in the price of solar PV modules in the last decade, [2] there is a lot of interest in diversifying the supply chain for solar PV technologies to increase domestic manufacturing and reduce costs even further.Since 2009, perovskite solar cell (PSC) technology has attracted attention in the PV research community as a potentially ultra-low-cost, high-efficiency thin-film photovoltaic (PV) technology. [3]Within a little more than a decade, PSCs have attained a power conversion efficiency (PCE) similar to silicon solar cell (SCs), exceeding the 25.0% mark in 2021. [4]he term "perovskite" refers to a class of materials having a particular crystal structure and the general stoichiometry ABX 3 , where A site is a large monovalent inorganic (most common caesium) or small organic cations like methylammonium or formamidinium, B site is a divalent metal cation and X stands for a halide anion. [5]Since perovskite PV can be produced with common materials, using much less energy and solution processing methods, researchers are hopeful that this technology could deliver efficient solar energy at a fraction of the cost of existing technologies.The reasons behind the rapid increase in perovskite cell efficiencies can be attributed to the tuneable bandgap, high absorption coefficient, long carrier diffusion length and remarkable electrical properties. [6]owever, there are many problems to solve before perovskite PV modules can be installed in the field.Upscaling lab-scale cells into modules is one of the challenges.Currently, there is a very large difference between the efficiency records for PSCs, which are typically <0.1 cm 2 and mini-modules, which are larger than about 10 cm 2 and far less efficient.The energy production of the global solar module markets has increased from 867 MW in 2004 to 177.7 GW in 2021 and is expected to reach 536.9 GW in 2027 and 1780 GW by 2050. [7]This growth has largely occurred with c-Si and CdTe module technologies.To commercialize perovskite solar technology, at least three key challenges need to be addressed: 1) reduce the cell to module efficiency losses while increasing the size of modules produced; 2) develop rapid and accurate module characterization methods for this technology; and 3) significantly increase the operational lifetime of modules.
To address (1), significant efforts have already started to make perovskite modules larger.For industrial applications, module sizes are commonly divided into four categories with sizes ranging from 200 to over 14 000 cm 2 including mini-module, smallmodule, standard-module, and large-module. [8]One of the largest perovskite solar modules with an effective area of 1241 cm 2 has been introduced by Suzhou GCL Nano Technology Co., Ltd., but it just barely touches the bottom of the small-module size in general. [9]hallenge-( 2) is the difficulty of measuring the performance and efficiency of a perovskite module.Since PSCs suffer from hysteresis and/or meta-stability, accurate measurements require the devices to arrive at their stabilized status.Thus a continuous solar simulator along with precise temperature control is needed, which consumes lots of time.The flash simulator used for the rapid screening of the c-Si modules may not work for perovskite modules.Moreover, significant differences in measuring procedures and reporting guidelines between labs have remained a big issue for statistical analysis of the stability data over the past ten years.
The challenge (3) still persists for both lab-scale PSCs and modules.The commercialized monocrystalline solar modules can operate at 95% (T 95 ) of their initial efficiency after 10 years in the field and at 90% (T 90 ) after 20 years, and the system is expected to last for up to 30-40 years or more. [10]Obtaining such lifetimes presents a substantial challenge for perovskite solar modules due to the materials' inherent instability to environmental factors such as light, moisture, oxygen, and heat.
In this review, we discuss recent progress in perovskite minimodule development focusing on scaling up the module area while reducing cell-to-module losses, a review of methods to measure the performance characteristics of modules (including hysteresis or short-term metastabilities) and efforts to measure and increase module lifetimes.Based on a literature survey of 70 papers reporting perovskite mini-module performance from 2015 to the present, we have reviewed progress in these three challenge areas and provided suggestions for future research priorities that support the eventual commercialization of this technology.While an efficiency loss (ΔPCE) when scaling the lab scale cells to minimodules is expected, the main causes of the loss remain elusive.Therefore, we analyzed the relation of the ΔPCE with the area of the mini-module, and does not necessarily increase with the area of the module.Further, we found that coating methods play a big role in module efficiency upon scaling up due to variations in the film quality, uniformity, and/or morphology.If the perovskite coating quality is well maintained when upscaling, we project a >24% PCE from a perovskite module based on the champion labscale cell efficiency.T 80 lifetime is reported over 1000 h in an inert environment, but the measurement protocols are not consistent between reports.The lifetime of perovskite modules is affected by intrinsic and extrinsic factors.With further improvement of encapsulating technology, device refinement, as well as new material development and module stability, could dramatically increase and meet commercial standards in the coming time.Existing performance and accelerated stress test protocols used for c-Si PV modules may not be appropriate for perovskite PV due to its slower and metastable device response and distinct failure and degradation modes.

The Current Status of Perovskite Mini-Module Development
A solar module consists of a series of SCs that are electrically interconnected and packaged to survive the operational environment.Figure 1a sketches a typical structure of a perovskite minimodule.Three steps are usually employed to connect the subcells into the module: first, the transparent conductive electrode (TCE) is patterned (P1) with a laser scribe.Next, a selected bottom transport interface layer, the perovskite photo-absorbing layer and the top transport interface layers are coated on the patterned substrate.Then, the device is scribed again (P2) to isolate the photoactive structures and create an opening to connect with the transparent conductive electrode (TCE) layer.Finally, a top electrode such as gold, copper, or carbon is deposited and the top metal layer is scribed again to complete the module.The active region that produces electricity is confined between P1 and P3 while the rest of the area is considered the dead area.Figure 1b summarizes commonly developed module active areas reported in the literature.The vast majority of perovskite mini-modules recorded in the literature are 100 cm 2 or smaller for proof-ofconcept demonstrations.Larger modules of 200 and 300 cm 2 are reported by Yabing Qi and Hong Lin Groups, [11] respectively.In 2020, Panasonic Corporation reported an 802 cm 2 perovskite solar module with a PCE of 16.0% and later announced the certified PCE of 17.9% for a device with 804 cm 2 area, which sets a new record for the largest perovskite module in size. [12]Another company has produced larger modules of 1241 cm 2 (GCL Nano) but has not published their results nor gotten their modules certified at an independent lab. [9]igure 1c displays laboratory-scale cell-level PCE (blue) and the PCE of the modules as the cells are scaled up (red).Lab cell PCEs spread between 12-24% while module efficiencies are in the range of 7-20% with various perovskite compositions employed for module fabrications.Figure 1d plot reports module efficiencies with their publication year for three typical compositions: pure single organic cation perovskite (MAPbI 3 ), mixed-cation pure halide perovskite (Cs 0.5 FA 0.5 PbI 3 and MA 0.5 FA 0.5 PbI 3 ) and mixed-cation mixed-halide perovskite (MA 0.5 FA 0.5 PbBr 1.5 I 1.5 , MCMH) structures.Since 2014, reports were dedicated to MAPbI 3 -based mini-modules.Initial module efficiencies were around 8% evolving to 16-19% within recent years as MAPbI 3 perovskite technology continues to mature.MCMH-based modules gained attention starting in 2017 due to their potential for both efficiency and improved operational stability.MCMH module efficiency later rose from 16% to nearly 20%.The average efficiency of modules made with MAPbI 3 compared to those made with MCMH is displayed in Figure 2e.The latter has an average efficiency of 18%.

Cell-to-Module Losses
In general, the cell to module PCE losses for current commercial solar modules is greater than or equal to 0.8% in absolute terms. [13]This inverse scaling phenomenon has been observed in different types of thin-film SCs, including amorphous   [1b] The different line represents modules with different scribe widths (micrometers).b) Projected efficiency using the champion cell efficiency from a typical a-FAPbI 3 device [4] as a function of scribe width.Here the cell width is 6.5 mm (P1 to P1 distance).13a,14] Improving the electrode conductivity through composition and hybrid engineer-ing while maintaining transparency will be the key to achieving the ideal ITO electrode for efficient solar module fabrication.Recently, a design of silver mesh at an optimized aspect ratio on a transparent ITO electrode can reduce the sheet resistance below 1 Ω/□ and maintain the transmittance exceeding 85% of solar modules. [15]Furthermore, various ITO-alternative materials such as graphene, metal mesh, conductive polymers, metal nanowires, carbon, and nanotubes have also been developed for PV's transparent electrode to improve the conductivity upon scaling the device. [16]PSCs and modules show the same inverse scaling trend but they exhibit much higher losses from cell to module than other thin-film solar technologies (Figure 3b,c).It is not yet clear why these losses are much higher as no mechanism has yet been identified.To understand the key factors that affect perovskite module efficiencies, a detailed analysis comparing the reported IV characteristics of the SCs from their corresponding module is carried out in Figure 3. Analyzing each characteristic provides clues on which parameters play major roles in the PCE of the module.
Modules fabricated according to the layout in Figure 1a can be represented with an equivalent circuit diagram as shown in Figure 3a.Each sub-cell can be described by a power source, a series resistor (R S ), and a shunt resistor (R SH ) to describe the loss.All sub-cells are connected in series with a load resistor.

Open-Circuit Voltage (V OC )
The module's open-circuit voltage (V OC, module ) is the sum of the V oc of all the sub-cells (V OC, sub-cell) : In an ideal condition, all sub-cells output the same V OC denoted as V.The V OC, module equals the number of sub-cells (N) multiplied by Voc 0 .V OC loss originates from the energy loss via charge recombination either through electronic trap states or through shunt recombination near the interfaces.Perovskites are materials that are mostly immune to defects, as revealed by Petrozza et al. [17] Structural defects in perovskites do not create deep levels of electronic trap states.In real life, if R SH from all cells or one sub-cell isn't infinitely large, V OC, module will deviate from the summed total of V OC .Charge recombination through shunts can be significant in the presence of local pin-holes and/or film non-uniformities.Therefore, the main loss in the V OC originates from the coating quality of the perovskite layer, which includes the thickness uniformity, crystal size and film morphology.

Short-Circuit Current Density (J SC )
When short-circuiting the module, that is, set R load = 0, all subcells are not typically in short-circuit conditions.They are loaded by the neighboring sub-cells' R S .If the R S is very small (→0) , the J SC of the module (J SC, module ) is determined by the smallest short-circuit current (I SC, smallest ).If we assume all sub-cells have identical area and output current, then the module's J SC, module is: The J SC,module can be described by J 0 divided by N. If one subcell outputs a smaller I SC because of a smaller area or current loss through the series resistance (R S ), the total output current would be lower.Additionally, J SC has another loss pathway.Additional R S adds up the internal resistance which lowers the overall FF of the module, resulting in a lower current value when the bias departs from the SC point.A more complex situation could occur if the FF of individual sub-cells is non-uniformly distributed, then the J SC, module will be determined by the worst cell.

Fill Factor (FF)
The fill factor (FF) of the module represents the maximum power point (MPP) which is critical for the module's field tests.
In a typical field test, modules are tracked at their MPP by a perturb-and-observe algorithm that adjusts operating voltage to keep the device at its maximum power.Finding the MPP is crucial to optimizing the module's performance.For a reasonably performing module, one would not expect a significant loss in V OC nor J SC .Therefore, tracking the MPP depends highly on the change in the FF.
In Figure 3b (panel 1), we compare cell efficiencies to their corresponding module efficiencies which include their differences (ΔPCE = PCE cell − PCE module ) plotted in the 2nd panel.The data from left to right are arranged according to the modules' active area to visualize any trends.From Figure 3b, we do not find an obvious trend in the ΔPCE when the area scales up nor when the number of sub-cells increases.This indicates that efficiency loss in scaling lab-scale cells to modules is not directly tied to the module's size.In other words, scaling a cell to a module does not necessarily lower the output performance.To get a deeper insight into this loss, we compare the rest of the IV characteristics: V OC, module , J SC, module and FF, to analyze their trends with module area and number of sub-cells in Figure 3d-f.Figure 3d plots the V OC, module against the sub-cell values.According to Equation (1), this should follow a linear dependence and the slope should equal the average value of the small-scale cells.From this data, we obtain a slope value of V 0 = 0.99 V, which is 30 mV lower than the lowest V OC from the small-scale cells in the current data pool.This result is 115 mV away from the average V OC (1.105 V).The lower panel of Figure 3d shows V OC loss when comparing cells to modules that lie in the range of ±0.2 V. Interestingly, not all ΔV OC values are negative.Some modules have a higher average V OC than the reported small cell which is likely due to the wide distribution of achievable V OC from cell-tocell.
Figure 3e presents J SC, module against sub-cell values.Based on our analysis, J SC, module is determined by the smallest J SC from the sub-cell.Fitting the data yields a slope of J 0 that describes the J SC of the lowest delivering cell.The fitted J 0 from the curve is 19.5 mA cm −2 , which correlates to the average of the lowest J SC from the data pool.This value is 0.5 mA cm −2 smaller than the lowest performing lab-scale cell and is 2.9 mA cm −2 worse than the average J SC of the lab cells.The percentage of J SC drop, (ΔJ SC /J SC ), is shown in the lower panel of Figure 3e.The values are mostly negative in the range of 10-40%, suggesting the J SC from the module is generally lowered when scaled up.According to our previous analysis, the decrease in J SC is either determined by the lowest deliverable J SC from a sub-cell or current loss through R S .
Figure 3f is the FF of the reported modules.When the module area exceeds 100 cm 2 , FF drops below 70%, whereas, the FF values are in the range of 60-80% for smaller modules.This loss in FF can be attributed to both R S and R SH .Additionally, given that the V OC loss from the module is generally very moderate, the FF loss can be mainly accredited to R S .
In the PV industry, cell-to-module efficiency loss is often expected and mainly attributed to two factors: one is the current loss through R S added by the lateral resistivity of the transparent conductive oxide layer in large-scale devices; another is the dead area loss (e.g., the gap between P1 to P3) and handling area near the edge of the module which occupies the illumination area without producing power.1b,4] Figure 2a demonstrates the simulated module efficiencies as a function of cell width, which includes the effects of the dead area from the P1-P3 scribing.In this simulation, we use the highest single-cell efficiency of 25.59% and the TCO conductivity was chosen to be 7 Ω/□.The P1-P3 scribe dead area (SDA) varied from 150 to 520 μm.The active cell area is 1826 mm 2 with a fixed width and length of 32.55 and 56.1 mm, respectively.The laser scribe runs parallel to the length of the module.It is apparent that SDA and cell width have a significant impact on module efficiency.When SDA is small, module efficiency is highest, and the ideal cell width changes with SDA.The optimal efficiency was achieved at ≈5 mm cell width and 150 μm of SDA at 24.45%.Within the same model, we can project the module's efficiency by investigating the most-efficient lab-scale PSC with an alpha-FAPbI 3 as the solar absorber (assuming it could be replicated at a larger scale).A TCO conductivity of 7 Ω/□ is selected here and the efficiency is plotted in Figure 2b as a function of P1-P3 scribe widths.For instance, with a 150 μm dead area from the scribe, the module efficiency is ≈24.2%.When the dead area width increases to 500 μm, the projected module's efficiency is still above 23.0%,2.5% below the cell efficiency.Based on the literature data presented in Figure 3, a FF loss when the module area exceeds 100 cm 2 is indeed observed to be likely attributed to the TCO conductivity loss.However, this does not account for the efficiency loss for smaller modules.From our literature review, we also noticed that the reported module's efficiencies are calculated using the "active" area of the module.22b] Copyright 2018, Springer Nature Publishing.
has been already removed from the efficiency loss.The procedures established in research laboratories for the fabrication of small-scale cells are not suited for the large-scale manufacturing of modules.This point is often ignored when interpreting the performance between SCs and modules.
Considering this factor, we continuously analyze the efficiency of perovskite solar modules with a variety of fabrication strategies.Figure 4a-c presents common scale-up deposition methods that have been used to fabricate perovskite films such as spin-coating, slot-die coating, chemical vapor deposition (CVD), and blading coating. [18]3a] However, it is limited to a lab-scale cell or mini-module fabrication, and is not suitable for large-area panel manufacturing due to the poor uniformity from center to the edge.And it is a challenge to spread the liquid over a large area to achieve a full coverage.Therefore, a large amount of solution is required to initiate the spin coating process, which results in a significant amount of perovskite precursors (around 95%) being wasted during the spinning process.Slot-die coating and blade coating, in contrast, offer several advantages over traditional methods like spin coating.These techniques enable large-area coating with minimal solution waste and can be integrated into scale-up processes, including roll-to-roll coating and sheet-to-sheet deposition systems.Nonetheless, the high roughness and thickness variations of slotdie-coated and blade-coated films may present some drawbacks for SC devices. [19]CVD offers several benefits, including the creation of uniform films with minimal porosity, exceptional purity, and stability. [20]While the CVD of perovskite is commonly used in lab-scale SCs, it also has some drawbacks, such as the need for costly equipment and the release of harmful gaseous byproducts during the reaction.A comprehensive comparison exhibits that perovskite solar modules fabricated by the spin-coating method resulted in much lower PCE (≈6%) than small-size cells, which had a PCE of 8.6% and 15.4%, respectively.In contrast, the CVD of perovskite modules (PCE of around 15.6%) exhibited less loss (<1%) as compared to their small-scale counterparts (PCE of ≈14.7%). [21]We also found that nitrogen knives accompanied by slot die or blade coating yield champion efficient modules (with a decrease of 1-2%) among solution processing methods (Figure 4d). [22]The PCEs of cell and module were found to be 19.7% versus 15.8% and 17.2% versus 15.1% for blade and slotdie coating, respectively.These results indicate that the PCE gap between small-scale cells and larger-area modules might originate from the difference in quality, uniformity, or morphology of the perovskite film made by different deposition methods.

Challenges of Perovskite Module Characterization
Unlike conventional c-Si PV technology, which has very minor performance stability issues (e.g., light-induced degradation and light and elevated temperature degradation), perovskite PV appears to be much less stable and measuring the efficiency of PSCs or modules can be challenging and require special equipment, such as continuous solar simulators.Degradation and hysteresis (known as short-term metastability) are recognized as two critical issues influencing the stability and accuracy of perovskite PV measurements.In this section, we will discuss factors responsible for hysteresis and long-term instability in recently developed solar modules and review current efforts to reduce these effects.

Origin of Current-Voltage Hysteresis
It is known that PSCs exhibit current-voltage (J-V) hysteresis, for example, J-V curves do not overlay when swept in forward and reverse directions or at different scan rates. [23]This poses a challenge for measuring the actual output efficiency.23a,25] A general agreement in the literature is that hysteresis behavior is more severe when the perovskite material quality is poor or the perovskite material has degraded.Consequently, ion migration or ferroelectric polarization is assisted by local defects near the grain boundaries or at the interfaces limiting device performance. [26]ver the past few years, the hysteresis mechanism and its minimization or elimination have become a primary focus of many researchers. [27]23c] The mechanisms of the hysteresis behavior and meta-stability of the lab-scale cells are well established in the literature; [28] however, it is not clear if the module's J-V curves replicate small cells' properties.To quantify the degree of J-V hysteresis, the following section will summarize the perovskite PV hysteresis behavior of modules compared to single cells expressed in the literature.

Hysteresis in Mini-Modules and Efforts at Reduction
Figure 5a (top panel) shows the HI values of the lab-scale cells (blue solid circles) and those obtained from the module's J-V curves (red solid squares).The cells' HIs are mostly below 0.2, suggesting a less than 20% difference when sweeping the J-V curves in different directions.The HIs for the modules are either comparable to the cells' HI or are more pronounced.A side-byside comparison of the difference in HI, defined as HI module − HI small cell (ΔHI, green hollow squares), and the corresponding module area (red hollow circles) is also presented in Figure 5a.
The ΔHI mostly populates in the range of 0-0.1, and some peak values greater than 0.1 are observed.These results indicate that once made into a module, the HI can get worse compared to the cell.To analyze this in detail, we performed a case study by picking representative high HI and low HI modules (Figure 5a). Figure 5b shows the J-V characteristics of a large HI module and its corresponding lab-scale cell studied by Tong et al.The RS and FS J-V curves obtained from the lab-scale cell (active area of 0.09 cm 2 ) yield a low HI of 0.07. [29]When this device is scaled to a larger-area module with active areas of 22.4 and 91.8 cm 2 , the HI values increase to 0.12 and 0.32, respectively.Interestingly, the hysteresis caused a discrepancy in the module J-V curve's V OC and FF, whereas the J SC remains reproducible.Figure 5c demonstrates a small HI case taken from the work performed by Bu et al., in which a high-performance hysteresisfree module was realized via additive engineering. [30]In this work, a high-quality FA 0.83 Cs 0.17 PbI 3 film was first fabricated yielding a high PCE of 23.02% in the forward J-V direction, but the PCE dropped to 20.71% in the reverse direction, resulting in a relatively large HI of 0.1.To overcome this problem, they added an ionic liquid additive, KPF 6 , into the precursor.Impressively, the J-V hysteresis was eliminated from both the lab-scale cell and the large-area module.A module with 65.0 cm 2 active area was demonstrated using this ionic liquid precursor and fabricated with an air-knife-assisted slot-die method.The module's PCE was 19.54% in the reverse scan and 19.22% in the forward scan.
In another recent report shown in Figure 5d, additive engineering and modified large-area deposition techniques were applied to achieve high-efficiency modules.1b] The photo-active layer consists of a mixed cation, pure iodide structure, namely MA 0.6 FA 0.4 PbI 3 and a solid-state lead-coordinating additive of carbohydrazide (CBH) was incorporated in the precursor.Two modules are demonstrated with active areas of 17.9 and 50.1 cm 2 .The PCE values determined from the J-V scan were 20.1% and 19.7% for the small and large modules, respectively.Notably, the efficiencies obtained from the J-V scans were very close to the certified values acquired with a stabilized, MPP tracking method.The certified PCE was obtained by tracking several current points near the MPP divided by the input sunlight power.As a result, 19.3% for 18.1 cm 2 module and 19.2% for the 50.0 cm 2 module were obtained, which were within 1% error compared to the J-V scans.This is attributed to the low HI of the module.
Because the hysteresis is tied to a slow transient behavior, next, we reviewed the time required for a module to reach its steady state.Figure 6a summarizes the reported stabilization times of the perovskite modules when held at their MPP under 1-Sun illumination, however exactly how these times are defined is not clear.Most of the modules in the literature pool can stabilize quickly within 50 s while some modules experienced slower stabilizations of over 100 s.We chose several typical maximum power stabilization curves to investigate in detail.
Upon changing states, such as rapidly changing irradiance or applying a bias on the module to find the MPP, the photocurrent of the perovskite module could either increase or rapidly drop before reaching its steady state.Figure 6b summarizes the number of cases observed in the reviewed studies. .HI analysis for the reported perovskite solar modules.a) Summary of the HI differences between the lab-scale devices and the corresponding modules.The data points are presented in a manner of increasing device area.Modules with active areas > 50 cm 2 are highlighted with a yellow background.b) RS and FS J-V curves obtained from the spin-coating devices with active areas ranging from 0.09 to 91.8 cm 2 .c) RS and FS J-V curves obtained from the slot-die printing devices with active areas ranging from 0.148 to 65.0 cm 2 .d) J-V curves obtained from the blade-coating devices with active areas ranging from 0.07 to 50.1 cm 2 .Stabilized module PCE values certified at National Renewable Energy Laboratory are also presented.Figure 5b is adapted with permission. [29]Copyright 2021, John Wiley & Sons.1b,30] Copyright 2021, American Association for the Advancement of Science.6c,d is adapted with permission. [30]Copyright 2021, American Association for the Advancement of Science. Figure 6e,f is adapted with permission. [32]Copyright 2018, Elsevier.
Bu et al., demonstrated a high-performance perovskite module through defect passivation via additive engineering. [30]This module stabilized almost instantaneously (pink line, Figure 6c) and the corresponding J-V curves were free from a hysteresis (pink solid and dash lines, Figure 6d).In contrast, the module without additive treatment exhibited a metastable power output behavior (blue line, Figure 6c) with a high HI of 0.19 (blue solid and dash lines, Figure 6d).These results validate the importance of passivating the defect to eliminate metastable power outputs and J-V hysteresis.In another work reported by Z. Liu et al., an interface stabilization method, which involved the engineering of all the relevant interfaces in a perovskite module, was used to identify high-performance modules with excellent stabilities. [31]iven the high stabilized PCE value, a gradual increase in power output is observed during the first 30 s of the SPO measurement.The corresponding J-V curves yielded a mild HI of 0.09. [31]An example showing decreasing power output during an SPO measurement was found in a published report by E. Calabrò et al. [32] As shown in Figure 6e, the power output of the module experienced a significant drop before reaching a steady state.In addition, a large HI of 0.18 was determined according to the J-V curves (Figure 6f).These results suggest metastability needs to be characterized well so that a valid PCE measurement can be made.
In presence of a hysteresis or short-term meta-stability effect, it becomes challenging to accurately measure the module's efficiency with conventional methods performed during a field test.Therefore, pre-conditioning protocols aim for a steady-state pho-tocurrent before device characteristics are measured.However, pre-conditioning takes time and it is impossible to pre-condition every production module before field testing.In addition, if the metastable behavior occurs every day with the day-night cycle, this behavior must be understood so that accurate performance estimates can be made.For an indoor test, researchers typically characterize modules by measuring J-V curves at STC to represent power.However, field testing depends on many other environmental stress factors that influence module performance.Ideally, the measured current of metastable devices will be dependent on the measuring conditions (voltage, temperature, irradiance, environmental conditions) and not the prior history of the device.However, perovskite response times (seconds to minutes) require longer sweep time and scan rates to allow for sufficient current stabilization at each voltage step during a J-V curve.Standard testing methods need to be developed and implemented to evaluate the lifetime and stability of metastable devices without special treatment to stabilize the current before performance characterization.A PSC performance calibration guidance has been established by researchers from NREL addressing the complex issue involving the dynamic behaviors in the current-voltage characteristics. [33]

Increasing Module Lifetime
Commercial PV modules meant for long-term energy production are required to last at least 25-30 years and sometimes more without significant degradation.For example, First Solar's Series 6 CdTe modules have a 30-year linear performance warranty with an annual degradation rate of no more than 0.3% per year.This means that by year 30 the module power rating should not be below 91% of the original power.Even the most stable perovskite PV modules have not come close to reaching these industry expectations.While there may be a business case for a much less expensive PV module with a shorter lifetime, current efforts are largely focused on how to make perovskite modules last as long as conventional PV technologies.
Despite the unprecedented progress in achieving higher PCE, the short operational lifetime of perovskite PV is a major challenge that needs to be addressed for the commercialization of perovskite PV modules.There are two ways to evaluate operational lifetimes: 1) operate the module outdoors under load (e.g., MPP tracking) or 2) expose the module to accelerated testing indoors.The advantage of field testing is that the module is exposed to typical operational environments.Accelerated testing has the advantage of being repeatable, however, there is a risk that exposing modules to sequential stresses (e.g., thermal cycling, UV, light and elevated temperature, damp heat, and humidity-freeze) may not replicate realistic outdoor conditions, resulting in either failure that would not be seen in the field or a lack of failures that are seen in the field.Another problem with accelerated testing for perovskites is few standards to ensure that different research groups are testing their modules under the same conditions.Recent efforts to provide testing guidelines have been published for cells and modules. [34]Our review of the literature shows that few research groups test their cells and modules in the same way.The long-term operational stability (defined as the time for the output power to decline to 80% of the initial rated power) of a perovskite mini-module has been reported for about one year. [35]A very recent breakthrough demonstrated a 0.5 m 2 perovskite solar panel had PCE of 16.4% and 14.3% for reverse and forward scans at 1 sun irradiation and a remarkable T 80 of 5832 h in outdoor characterizations. [36]The high stability of the module was attributed to the use of 2D materials (MoS 2 and graphene) and Kapton foil which stabilized module interfaces and protect the device from environmental factors.However, the long-term stability of current perovskite modules is still far away from the 25-30 years expected for commercial solar panels.
The instability of perovskite solar modules is assigned from both intrinsic and extrinsic instability issues.Intrinsically, even if the cells are well-protected, performance degradation could still occur.This is attributed to ion migration and decomposition of perovskite materials.Extrinsically, if the perovskite devices are exposed to ambient environments; temperature, light moisture, and oxygen can damage the device.Moisture is proven to decompose the structure via an electrochemical reaction which can be accelerated by photo-excitations.Oxygen can diffuse into the perovskite layer, oxidize the material, and form lead oxide in the structure.Both the intrinsic and extrinsic degradation processes are initiated by defects and interfaces that hinder the lifetime of the device.This can be further improved by eliminating defects and grain boundaries while protecting the interfaces during fabrication.Extrinsic degradation can be avoided by developing a robust encapsulation strategy capable of isolating the thin-film layers from the atmosphere.

Intrinsic Stability
In principle, the failure of perovskite PV to maintain its maximum performance over a prolonged period may due to the deterioration of the light-harvester materials due to chemical or structural changes.One study found that perovskite materials tend to decompose into their corresponding components by the following pathway: [37] In a typical example, the density functional theory calculations of CH 3 NH 3 PbI 3 showed that the reaction enthalpy was relatively low (−0.1 eV) at room temperature, suggesting unwanted decomposition leads to the low stability of the perovskite compound at room temperature.Reaction enthalpy grew to −0.25 eV, and −0.7 eV when replacing iodine with bromine and chlorine, the materials were more stable and spontaneous phase separation did not occur. [38]The Equation (6) depends on whether the decomposition occurs under thermodynamic or illumination control.37b] In contrast to conventional photo-absorber materials like silicon, metal halide perovskite materials exhibit substantial ionic properties which limit their long-term stability due to the low activation energy for ion migration within the active layer. [39]In the absence of external factors (i.e., moisture and heat) ion migration in perovskites has been shown to induce the IV hysteresis, degrade perovskite structure, change the local bandgap of perovskites, and interact with electron or hole transporting materials. [40]Consequently, the devices suffer from decreased PCE during operation or when stored.It has been noted that when a device is subjected to the photo-illumination external electric bias or thermal stress, the ion diffusion can increase.Therefore, numerous studies have been performed to reduce ion migration and improve the structural stability of the materials including low dimensional perovskite development, additive engineering, composition engineering, and grain boundary passivation to improve the long-term performance of the perovskite modules. [41]

Extrinsic Stability
After the intrinsic stability of the PVs has been addressed, the stability of the cells against external factors such as moisture, oxygen (air), heat, light, external bias, etc. will be crucial.It is important to note that external conditions can accelerate the reversible or irreversible deterioration of perovskite PV modules. [42]In addition, the modules have more interfaces than a cell near the interconnection sites, which may induce more failures from environmental factors than the small cells.Moisture, both alone and in the presence of oxygen, is one of the most widespread variables impacting the long-term stability of PVs. [43]Due to the highly hydrophilic characteristics, perovskite materials can easily absorb water molecules to form monohydrate or dehydrate compounds, which can change the characteristics of perovskite locally. [44] 3 NH 3 PbI 3 (s) These reactions are reversible.For example, if the hydrated products are stored in an inert environment, they can be dehydrated to regenerate perovskite with slight irreversibility due to phase segregation.However, when the moisture has saturated the perovskite, further water ingression or illumination of the materials might result in irreversible structural degradation Extensive studies explored that the reaction between perovskite and oxygen alone was thermodynamically unfavorable.However, simultaneously exposure of the devices to oxygen and light yielded faster degradation of their performance. [45]Under illumination, photoexcited electrons (*) in the perovskite interacted with oxygen molecules to generate superoxide (O − 2 ).This extremely active (O − 2 ) ion breaks down the perovskite structure by reacting with the organic MA. [46] This led to the rapid decomposing of the photoactive layer into precursor products as shown in the following reaction: Proper encapsulation of the perovskite devices has been shown to prevent degradation from moisture and oxygen. [47]Many efforts have been made to enhance the moisture and oxygen resistance of the perovskite materials, such as the passivation of surface perovskite with small molecules, [48] the introduction of a hydrophobic polymer layer, [49] the incorporation of less acidic cations and hybrid of 2D/3D perovskite materials. [35]Promising long-term stability was reported for polymer-coated devices tested on a roof that withstood rain and variable temperature conditions for more than 90 days, showing no sign of degradation. [50]ther extrinsic deterioration mechanisms have been connected to high-temperature device testing or devices heated by continuous light.During operation, solar modules will be subjected to elevated temperatures (up to 85 °C) in accordance with international standard measurements [such as IEC 61215 (2005), IEC 60904 (2006), IEC 61646 (2008), IEC 60891 edition 2.0 (2009), and IEC 61853 (2011)]. [51]However, phase change among orthorhombic, tetragonal, and cubic occurred when raising the temperature to above 25 °C.Conings et al. found that perovskite tended to decompose into PbI 2 at 85 °C within 24 h by heating the perovskite films in ambient air, oxygen, and even in nitrogen environments. [52]Similarly, Bertrand et al. investigated the thermal stability of CH 3 CH 2 PbI 3 at different temperatures ranging from 25 to 200 °C. [53]They observed that perovskite materials already decomposed into Pbl 2 at 100 °C in 20 min.So far, the decomposition of CH 3 CH 2 PbI 3 perovskite has been reported in a range of 85-140 °C and the decomposition process can be depicted as: It was believed that the MA cation was thermally unstable in perovskite and is entirely responsible for deterioration. [54]Furthermore, several investigations have now demonstrated that spiro-OMeTAD was crystallized at 100 °C that allows the interaction between the metal electrode and the perovskite materials, degrading the SC efficiency. [55]The use of carbon electrodes can release this problem because carbon materials are thermally stable and do not react with the active layer. [56]Using modified perovskites is currently the most suitable option to prevent thermal decomposition in perovskite devices. [57]An introduction of monoammonium zinc porphyrin (ZnP) into the perovskite film of the module offered enhanced thermal stability. [58]At a humidity of 40% and temperature of 85 °C, the mini-module could maintain 90% of the initial efficiency after 1000 h while a pristine perovskite-based device exhibited a loss of 95% of its performance within 250 h.The high stability of the module can be assigned for the attachment of ZnP units on the surface of perovskite crystallines which prevents the escape of cation and reduces defect generation, resulting in an effective molecular encapsulation and surface passivation of perovskite surfaces.
In real-world settings, SCs may be partially due to variations in object coverage, such as shades from flying birds or thick clouds.In such cases, shaded cells can be compelled to operate in reverse bias in order to match the flow of current through the rest of the module. [59]The breakdown voltage of the single cell has been identified to be at around −3.6 V. [60] If the reverse bias on the shaded cell exceeds its breakdown voltage, it begins to conduct electricity, generating heat through Joule heating. [61]This power is dissipated through low-resistance pathways (i.e., existing defects), leading to the creation of hot spots.These effects may couple with intrinsic and extrinsic ionic conduction phenomena, which can further reduce the overall performance and shorten the lifespan of the module. [62]The use of a thermally conductive substrate in substrate-oriented modules can benefit the distribution of the heat generated more evenly across the entire cell, effectively reducing the maximum temperature of hot spots. [61]o combat the degradation caused by reverse bias, it is possible to address the slow decay process at the cellular level.This process is likely due to an electrochemical reaction involving ionic species at the interface between the perovskite and the contact.By introducing an ion-blocking layer between these two materials, this degradation can be mitigated. [63]This intervention leads to an increase in the reverse bias current, which then stabilizes, allowing each SC to effectively function as its own bypass diode, preventing damage.Another approach for preventing degradation resulting from shading is to incorporate bypass diodes into the module.Figure 7a depicts the integration of a bypass diode in reverse polarity to a PSC, with both components separated by a lateral isolation region. [64]This implication will add a series resistance R s and a shunt R sh originating from the isolation region and the diodes to the cell circuit.The minimal fraction loss can be estimated as:   7 is reproduced with permission. [64]Copyright 2022, Elsevier.
Where , d and L, are the absorber resistivity, thickness and length of the cells.The acceptable range of FL should be below 10%, this value can drop further if the value of  is reduced (Figure 7b). [64]These findings imply that integral bypass diodes are eligible for any practical size of current PCs, and the loss of FL can be managed by monitoring .Incorporating integral bypass that were built vertically along the length of the cell or horizontally across the width of the cell can significantly reduce heating compared to modules without integral bypass diodes, as shown in Figure 7c.The vertical bypass module was entirely heated, while the horizontal-bypass module experienced more localized heating with a higher central temperature due to differences in the covered dimensions of the two configurations.As the cell length was reduced in the horizontal-bypass configuration, heating became more gradually coupled, affecting the main cell regions (Figure 7d).Therefore, the horizontal-bypass configuration, which prevents main cell regions from heating, provides better protection against overheating, as long as the bypass regions are not susceptible to local overheating.However, this installation of bypass diodes would increase fabrication costs.Herein, a larger breakdown voltage would be required to minimize the number of diodes.

Stability of Charge Transport Layers
Both electron and hole transport layers play crucial roles in facilitating efficient charge extraction and transport within cells, mitigating the occurrence of undesired charge recombinations, and ultimately leading to exceptional PV performance. [65]Additionally, charge-transporting layers act as a barrier that prevents external moisture or oxygen from reaching the perovskite materials.Therefore, the choice of charge-transporting materials is critical in influencing the efficiency and stability of perovskite solar modules.Recent studies have shown that the use of metal oxides, conducting polymer, and small organic molecules as charge transport layers can lead to high device performance.
Although TiO 2 and fullerene-based molecules have been identified as the most effective electron transport layer (ETL) material for PSCs, their susceptibility to oxidation under UV light and exposure to oxygen can catalyze the oxidative degradation of other materials in contact with it, thereby potentially reducing the device stability.Doping has been claimed to be an effective method for changing the electrical characteristics of electron transport materials and improving the stability of the devices. [66]he introduction of Al (2.5%) into TiO 2 could eliminate oxygen defects in the lattice structure of the ETL, resulting in increased performance and greater stability. [67]Sn-doped TiO 2 effectively passivated trap states, upward shifted conduction band minimum and valence band maximum, and improved the carrier extraction of TiO 2 .The perovskite device with Sn-doped TiO 2 electron extraction layer showed an enhancement of the PCE and stability by 4% and 50% compared to pure TiO 2 . [68]hen et al. introduced Nb 5+ ions into the TiO x matrix to improve rapid carrier extraction by reducing the density of pinholes and cracks over large areas. [69]As the result, the use of a Ti(Nb)O x layer to protect the perovskite from moisture intrusion enabled the module to maintain 97% of its initial PCE even after 1000 h of light exposure.Snaith and colleagues fabricated MCMH PSCs using an n-doping electron-accepting layer (C60).The wettability and conductivity of ETL films were improved by using the 4-(1,3-dimethyl-2,3-dihydro-1H-benzimidazol-2-yl)-N,N-diphenylaniline (N-DPBI) dopant.The perovskite devices exhibited a PCE of 18.3% and improved device stability with T 80 over 3600 h under full sun illumination in ambient conditions because the number of surface defects was reduced and trap states with higher electron density were filled. [70]Other efficient ETL materials have also been reported, including SnO 2 , ZnO 2 , SrTiO 3 , BaSnO 3 , Zn 2 SnO 4 , etc. [71] Spiro-MeOTAD is a frequently utilized hole transport layer in both traditional and mesoporous SCs, and has been shown to achieve a certified PCE of up to 23.7% in lab-scale PSCs. [72]To enhance the conductivity and hole mobility of Spiro-OMeTAD, it is commonly necessary to introduce doping agents such as lithium trifluoromethanesulfonimide (Li-TFSI) and 4-tert-butylpyridine (TBP). [73]However, the use of these dopants can lead to oxidation and degradation of the spiro-OMeTAD layers and the perovskite film upon exposure to ambient air. [74]Furthermore, the high cost of spiro-OMeTAD material presents a significant challenge to the development of PSCs for industrialization.To surmount these predicaments, a plethora of contemporary research has centered on utilizing low-cost conducting polymers to heighten the constancy of the hole transport layer.Among these materials, PE-DOT:PSS has been the most commonly used due to its commercial availability in a suspension form and its ability to provide a smooth and hydrophilic surface for perovskite deposition. [75]owever, the hygroscopic and acidic properties of PEDOT:PSS can deteriorate the stability of the hole transport layer and negatively impact the PV efficiency of PSCs. [76]Alternatives materials, such as poly[bis(4-phenyl)(2,4,6-trimethylphenyl)amine (PTAA), graphene oxide, molybdenum oxide, copper(I) thiocyanate and (CuSCN) poly(3-hexylthiophene-2,5-diyl) (P3HT), have been employed to replace PEDOT:PSS, and have shown significant improvements in both stability and performance. [77]

Progress in Improving the Module Stability
Among the 70 papers we have studied, about 30 investigated the modules' operational lifetime, but few of these studies were conducted using the same environmental stress conditions, which raises important concerns about our ability to compare results.Some tests are performed in an inert environment (nitrogen or with encapsulation), and some apply light and thermal stresses, although inconsistently to assess the intrinsic lifetime of the material.Additionally, most lifetime tests were performed by tracking the MPP of the module.It is also worth noting that most of the lifetime tests are done in laboratory testing conditions where a 1-Sun equivalent solar simulator constantly illuminates the modules with no dark and light cycles during the test.The temperatures are commonly kept at 25 °C in the initial stage of the measurements and the modules may reach 70 °C or more after several hours of operation. [79]Figure 8a plots the reported lifetime for the indoor tests.A T 80 quantity, for example, the time stamp when the PCE decays to 80% of the initial value, was commonly employed to describe the modules' lifetime.On the other hand, some studies tested their modules continuously for 1000 h and measured the lifetime of the PCE as (T X ), where X (X = 85, 95, etc.) was the percentage of the retained PCE of the device.In Figure 8a, one will notice that most of the modules among the reports we studied can maintain over 80% of their peak PCE over 1000 h.Such a T 80 lifetime was also achieved among the bestperforming lab-scale cells, suggesting the operational lifetimes were not impacted when made into modules.However, for modules to survive for 20-30 years they will have to survive much longer than 1000 h.
It is worth mentioning, the definition of the T X lifetime varies from paper to paper.An initial stabilization process reported by Liu et al. is not uncommon. [31]The first point, the stabilized point and the maximum point all affect the T X value and influence the normalization of the stabilization curve.For instance, both Figures 8b and 8c take the first point for normalization, whereas the maximum point (reported by Liu et al.) after the initial increase is considered for normalization.Some other reports employed the peak efficiency value to find their T 87 lifetime. [80]ne of the problems associated with the estimation of T 80 value is the "burn-in" and non-monotonic behaviors which are commonly observed while studying the long-term operation of perovskite solar modules. [81]Burn-in is typically defined as a rapid fall while nonmonotonic behavior presents a quick rise of the PCE in the initial stage, followed by a lengthier degradation period when the PCE declines more gradually (Figure 9a,b). [82]t is a prerequisite that these phenomena are often not observed under dark measurements, the origin of these behaviors is most likely related to external stress (electrical stress, thermal stress, or light soaking), material properties, layer stacks, impurities and blend morphologies. [83]It has been shown that the change of PCE at the initial stage is partially reversible and could originate from charge accumulation and imbalanced ion distribution. [83,84]81a] The large PCE variations cause difficulty to determine the real device's operational stability.To overcome such challenges, the original ISOS standard protocols recommended using the initial PCE value at the end of the burn-in area (t S ) to calculate T 80 for devices (named T S80 ) as indicated in Figure 9c,d. [85]ecently, modules produced by Liu et al. had a T 90 for 1570 h and a T 80 for 2680 h at a temperature of ≈40 °C. [31]This module had a PCE of 16.6% and an active area of 22.4 cm 2 .The stability was tested by constantly tracking the module's MPP under AM1.5G light illumination in a nitrogen box.Significant stability improvements are achieved via interface stabilization approaches, where ethylenediaminetetraacetic acid dipotassium salt (EDTAK), ethylammonium iodide/methylammonium iodide (EAI/MAI) and poly(3-hexylthiophene) (P3HT) were used to stabilize the interface among perovskite layer, charge transporting   8b is reproduced with permission. [35]Copyright 2017 & 2020, Springer Nature Publishing. Figure 8c is reproduced with permission. [78]Copyright 2021, American Association for the Advancement of Science.
layers and device encapsulation.Without proper interface passivation, the module tended to degrade much faster.Their demonstrations are consistent with the knowledge developed for perovskite cells where a clean interface is crucial for stabilizing the device.Protecting the interface has proven to be important.For example, Grancini et al. developed a perovskite solar module with a full year of stable performance. [35]The modules' absorber was MAPbI 3 and its surface was passivated by a 2D perovskite layer.The 2D surface passivation was key to achieving the long-term stability of a 10 by 10 cm module with an average PCE of 11%.The module was tested under constant 1-Sun illumination in a lab testing environment, and a temperature cycling up to 90 °C was also performed.Another work by Sha et al. employed an ultrathin interlayer of bridge-jointed graphene oxide nanosheets at the p-type interface to stabilize the module along with a dopantfree hole transporting material. [86]They achieved module stability of T 90 for 1000 h under continuous 1-Sun illumination at 60 °C in ambient air.
The external protection layer for the module has also proven to be critical to achieving a long lifetime.For example, Yang et al. demonstrated a protected perovskite module that has a T 95 for over 1200 h. [78]The authors employed MCMH composition, for example, FA 0.83 Cs 0.17 PbI 2.83 Br 0.17 , outputting a high PCE of 16.63%.By coating the module with an Al 2 O 3 barrier layer via atomic layer deposition along with cover glass encapsulation, the module retained 97% of its peak PCE for over 1200 h.In contrast, the module without the Al 2 O 3 barrier layer had a T 80 of fewer than 800 h.A few more papers studied the outdoor lifetime of their perovskite solar modules, and their efficiency data were taken periodically.In another work performed by Hu et al., a 2D/3D perovskite absorber was integrated into a 10 by 10 cm module. [87]Over 10% of total PCE was demonstrated, and more importantly, outdoor testing is conducted.This module was held at the MPP and the PV parameters were collected every 25 h.The efficiency of the module was maintained after 30 days in the field.
Until now, various encapsulation technique employing materials and structures that possess high barrier performance against oxygen and moisture was employed to protect these devices (Table 1).

Summary and Future Outlook
In this review, the developments and challenges for PV modules are discussed.The development of modules has shown an   8a,b is reproduced with permission. [83]Copyright 2018, Elsevier.unprecedentedly rapid improvement in PCE to values approaching 20% for single junction devices in less than 15 years.It was discovered that the superior efficiency performance of these devices was attributed to their exceptional optoelectronic characteristics such as tunable absorption, high defect tolerance, long charge carrier lifetimes and diffusion lengths, and absorption coefficient.At present, three main challenges exist before perovskite PV modules can be commercialized: 1) coating methods that maintain the high material quality when upscaling; 2) hysteresis, long-term operational stability; and 3) a unified device performance and lifetime characterization protocol.The largest perovskite module introduced by Suzhou GCL Nano Technology Co., Ltd. is ≈1241 cm 2 which belongs to the small module category.To be ready for solar PV plant applications, the size of perovskite modules needs to be at least 5-10 times larger.One problem associated with the scaling up of perovskite devices is the dramatic decrease of PCEs which is connected to the poor quality and uniformity of the perovskite film when applied over large areas.Recent research has shown that the CVD of the perovskite layer could help to maximize the performance of the perovskite module as compared to solution processing deposition methods.
The precise origin of meta-stability or hysteresis is ambiguous, but it has been shown that hysteresis is related to imbalanced charge carrier transport, trap-assisted charge recombination, the ferroelectric effect and ion migration.Various other factors such as device architecture, scan rate, scan direction, and voltage range also influence J-V hysteresis in perovskite modules.From prior research, by mitigating the ions' migration, a reasonable scan rate and a proper voltage pre-bias poling could really reduce hysteresis.It has also been discovered that crystal size and the composition of perovskite materials presented some influence on suppressing the hysteretic behavior.In addition, a standard measurement protocol needs to be developed in order to evaluate the performance of perovskite PV devices as accurately and unambiguously as possible.
Although the device's lifetime has increased from a few minutes to thousands of hours, this performance is still insufficient for commercial energy applications.The lifetime of the devices should be extended to more than ten years for potential commercialization.In general, burn-in is commonly characterized as the rapid increase or decrease of the PCE in the initial stage of longterm operational measurement due to charge accumulation and imbalanced ion distribution.This presents a challenge to accurately evaluate the device stability (e.g., T 80 ).Therefore, modified T 80 (named T S80 ), calculated from the initial PCE value at the end of the burn-in area (t S ) is recommended to use in determining the stability of perovskite modules as listed in the original ISOS standard protocols.Several degradation processes for perovskite PVs have been postulated, providing fundamental knowledge as well as some awareness for stability improvement.The inherent instability of the perovskite material itself is still under discussion, and further research should be engaged for a better understanding of its mechanism.External factors such as moisture, oxygen, light and thermal stress can accelerate the degradation of perovskite materials and thereby affecting the long-term performance of the modules.Several aspects of their systematic engineering, including structural design, charge transport materials, electrode material preparation and encapsulation techniques, must be considered to improve the stability of the perovskites.Advanced en-capsulation approaches might effectively improve the moisture and oxygen stability of perovskite devices.Both thermal and light stability issues of perovskite modules cannot be avoided during operation and their degradation mechanisms are still being debated.Furthermore, most researchers focus on measuring the long-term operation under constant illumination, other factors like temperature and packaging failures have not been studied, and require further investigation.Simply modifying the existing perovskite materials or interface will not be sufficient to achieve the big aim in terms of efficiency and stability; we advise inventing new materials and designs with high stability under adverse conditions.

Figure 1 .
Figure 1.a) Schematic illustration of a typical perovskite mini-module.b) Number of published papers versus the reported module area.c) Summary of cell efficiency (blue symbol) and module efficiency as a function of total active areas.d) Summary of the module efficiencies as a function of publication date for three main photo-absorbing perovskite compositions.e) Histogram of the module efficiencies fabricated with pure MAPbI 3 versus mixed-cation mixed-halide (MCMH) as the photo absorbers.

Figure 2 .
Figure 2. Module performance simulation.a) Simulated module efficiency using the module design as a function of cell width.[1b]The different line represents modules with different scribe widths (micrometers).b) Projected efficiency using the champion cell efficiency from a typical a-FAPbI 3 device[4] as a function of scribe width.Here the cell width is 6.5 mm (P1 to P1 distance).

Figure 3 .
Figure 3.Comparison of the IV characteristics of the lab scale perovskite cells and the modules reported in the literature.a) Equivalent circuit of a mini-module made with sub-cells connected in series.b) The PCE of mini-module and their corresponding small-scale cells compared in the same plot, the data points are arranged in an ascending manner based on their active area (number of sub-cells).The difference between the cell and module is plotted in green.c) The loss in PCE from cell-to-module distribution across the studied cases.d,e) Module V OC and J SC as a function of the number of sub-cells.f) The module's fill factor (FF) as a function of module area.

Figure 5
Figure 5. HI analysis for the reported perovskite solar modules.a) Summary of the HI differences between the lab-scale devices and the corresponding modules.The data points are presented in a manner of increasing device area.Modules with active areas > 50 cm 2 are highlighted with a yellow background.b) RS and FS J-V curves obtained from the spin-coating devices with active areas ranging from 0.09 to 91.8 cm 2 .c) RS and FS J-V curves obtained from the slot-die printing devices with active areas ranging from 0.148 to 65.0 cm 2 .d) J-V curves obtained from the blade-coating devices with active areas ranging from 0.07 to 50.1 cm 2 .Stabilized module PCE values certified at National Renewable Energy Laboratory are also presented.Figure5bis adapted with permission.[29]Copyright 2021, John Wiley & Sons.Figures5c and 5dare adapted with permission.[1b,30]Copyright 2021, American Association for the Advancement of Science.

Figure 6 .
Figure6.a) Distribution of the required times for the devices to reach their steady state.b) Several cases of the 3 types of metastable behavior observed in the reviewed literature.c) Steady-steady power output (SPO) and d) J-V curves acquired from the instantly stabilized module.e) SPO and f) J-V of the module demonstrate a decreasing power output during the metastable state.Figure6c,d is adapted with permission.[30]Copyright 2021, American Association for the Advancement of Science.Figure6e,f is adapted with permission.[32]Copyright 2018, Elsevier.
Figure6.a) Distribution of the required times for the devices to reach their steady state.b) Several cases of the 3 types of metastable behavior observed in the reviewed literature.c) Steady-steady power output (SPO) and d) J-V curves acquired from the instantly stabilized module.e) SPO and f) J-V of the module demonstrate a decreasing power output during the metastable state.Figure6c,d is adapted with permission.[30]Copyright 2021, American Association for the Advancement of Science.Figure6e,f is adapted with permission.[32]Copyright 2018, Elsevier.

Figure 7 .
Figure 7. Thermal reduction in perovskite solar modules under reverse biases using bypass diodes.a) Schematic illustrating the cross-section of a PSC integrated with a bypass diode.b) The fractional loss as a function of absorber resistivity in the PSCs with different cell lengths.c) The temperature rises under partial shade for three different module configurations: a standard module (left), a module with integral bypass diodes built vertically along cell length (center), and a module integrated horizontally across cell width (right).d) Temperature evolution of a module with integral bypass diodes built horizontally across cell width with various cell lengths.Figure7is reproduced with permission.[64]Copyright 2022, Elsevier.

Figure
Figure 7. Thermal reduction in perovskite solar modules under reverse biases using bypass diodes.a) Schematic illustrating the cross-section of a PSC integrated with a bypass diode.b) The fractional loss as a function of absorber resistivity in the PSCs with different cell lengths.c) The temperature rises under partial shade for three different module configurations: a standard module (left), a module with integral bypass diodes built vertically along cell length (center), and a module integrated horizontally across cell width (right).d) Temperature evolution of a module with integral bypass diodes built horizontally across cell width with various cell lengths.Figure7is reproduced with permission.[64]Copyright 2022, Elsevier.

Figure 8 .
Figure 8. Module stability characterizations.a) Summary of the reported operational lifetime against their T X .b) Typical module stability characterization of 2D/3D perovskite solar module at a temperature of 55 °C and short circuit conditions under 1 sun AM 1.5 G conditions.c) Operational stability of formamidinium-cesium (FACs) mixed-cations perovskites encapsulated by a cover glass or ALD Al 2 O 3 /cover glass under continuous 1-sun light at the temperature around 50 °C in ambient air.Figure8bis reproduced with permission.[35]Copyright 2017 & 2020, Springer Nature Publishing.Figure8cis reproduced with permission.[78]Copyright 2021, American Association for the Advancement of Science.

Figure
Figure 8. Module stability characterizations.a) Summary of the reported operational lifetime against their T X .b) Typical module stability characterization of 2D/3D perovskite solar module at a temperature of 55 °C and short circuit conditions under 1 sun AM 1.5 G conditions.c) Operational stability of formamidinium-cesium (FACs) mixed-cations perovskites encapsulated by a cover glass or ALD Al 2 O 3 /cover glass under continuous 1-sun light at the temperature around 50 °C in ambient air.Figure8bis reproduced with permission.[35]Copyright 2017 & 2020, Springer Nature Publishing.Figure8cis reproduced with permission.[78]Copyright 2021, American Association for the Advancement of Science.

Figure 9 .
Figure 9. Kinetic degradation in perovskite solar modules.a,b) Initial efficiency fluctuations in perovskite devices show a significant drop (a) or raise (b) in PCE.c,d) Schematic illustrating the common estimation of T 80 and modified T 80 (T S80) for perovskite solar modules presenting the burn-in (c) and nonmonotonic (d) behaviors.Figure8a,b is reproduced with permission.[83]Copyright 2018, Elsevier.
Figure 9. Kinetic degradation in perovskite solar modules.a,b) Initial efficiency fluctuations in perovskite devices show a significant drop (a) or raise (b) in PCE.c,d) Schematic illustrating the common estimation of T 80 and modified T 80 (T S80) for perovskite solar modules presenting the burn-in (c) and nonmonotonic (d) behaviors.Figure8a,b is reproduced with permission.[83]Copyright 2018, Elsevier.

Table 1 .
Encapsulated modules along with their stability.