Key bottlenecks and distinct contradictions in fast commercialization of perovskite solar cells

Despite significant improvements in photo-electricity conversion efficiency of perovskite solar cells (PSCs) over the past several years, this emerging photovoltaic technology is still years away from large-scale commercial application. In this review, important research progresses on PSCs’ ‘golden triangle’ parameters of efficiency, stability, and cost in literatures were objectively analyzed. We focused on their key bottlenecks and distinct contradictions hindering their fast commercialization. We also proposed the most urgent directions requiring intensive research and development input in the coming years to speed up the commercialization process of PSCs.


Introduction
As research has progressed, perovskite solar cells (PSCs) have established the basic groundwork for commercial application. That is why amounts of industrial capital were proactively invested in these two years. At this moment, it is essential to consider the following big things worthy of intensive input from research and development. Realistically, PSCs cannot yet compete with the mature crystalline silicon solar cells in the application scenario of ground power stations [1]. Regarding to the well-known 'golden triangles' of efficiency, stability, # These authors contributed equally to this work. * Author to whom any correspondence should be addressed.
Original content from this work may be used under the terms of the Creative Commons Attribution 4.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. and cost, PSCs may only offer doubtless advantage in cost after the initial stage of mass production (figure 1). New types of crystalline silicon solar modules, such as, those based on heterojunction technology and tunnel oxide passivating contacts (TopCon), will achieve mass production efficiencies of 23%-25%, with a lifetime of up to 20 years in the near future [2]. While PSCs currently have achieved record efficiency of 25.7% only for ∼0.1 cm 2 lab cells [3], while for minimodules with the size of ∼20 cm 2 , the efficiency record was 22.4%, and for ∼800 cm 2 submodules, the efficiency record declined to be 17.9% [4]. Achieving a mass-production module efficiency of over 22% with the comparable size of large silicon solar modules (>10 000 cm 2 ) is an impossible goal in the short term. In particular, the most efficient solar cells with >25% efficiencies rely on the use of unstable Spiro-OMeTAD hole transport materials (HTMs) [5] and noble Au electrodes [6], preventing them from mass production in industry. In terms of lifetime, only a few research institutes have reported the operational stability of PSCs without declining for 1000 h of

Future perspectives
Perovskite solar cell (PSC) is the most promising next-generation photovoltaic technology, which has a great potential to be commercialized in the near future. Its potential efficiency advantage and evident cost advantage are very clear. However, at the current stage, it is difficult to reconcile high efficiency (especially largearea high efficiency), stability and low cost. The light, thermal and moisture-resistance stability of devices based on different structures and key materials have many inconsistent performances, and even present contradictions. In view of these, it is very important to identify the key bottlenecks and the most critical technical obstacles, and dedicate effort in the application-oriented research directions for promoting the industrialization of PSC. continuous illumination and fewer with an illumination time of 10 000 h. Especially a lot of the reported operation stability has numerous specific preconditions (such as device temperature, the spectrum of the light source, applied potential bias, etc.), and there are various concerns with the reliability and reproducibility of the stability data. Up to now, only a small number of PSCs innovation companies, such as, Oxford PV, Co. in Germany and Microquanta Semiconductor, Co. in China, have reported their devices' stability results certified from third-party organizations. Substantial improvements in operation lifetime are in urgent need for today's PSCs. From above, we could identify the main bottlenecks preventing PSCs from fast commercialization are large-area high efficiency and longterm stability. Further detailed analysis can be considered for the following four contradictions to be solved.

The overlarge efficiency gap between small-area and large-area PSCs
As can be seen in figure 2(a), the efficiency of large-area perovskite solar modules (PSMs) is substantially lower than that of small-area devices, and it declines as the device area increases. PSCs have a far faster efficiency decline rate than other types of commercial thin-film solar cells (figure 2(b)), which occurs at a rate of around 0.8% in absolute value for each order of magnitude increase in the solar cell area. Maximum efficiency for a small area PSC is now 25.7%, whereas an 804 cm 2 sized PSM only achieves 17.9%. This overlarge efficiency gap implies a technological amplification challenge in PSCs, which needs to be overcome in the short term. Spin-coating methods [7][8][9][10], such as antisolvent extraction and surface modification/passivation, are typically used to coat small-area PSCs. These methods, based on spin-coating optimization, have been proved to be effective for high efficiency, which are difficult to be applied in fabricating largearea PSMs. Therefore, alternative coating techniques, such as, slot-die coating, ink-jet printing, etc, must be developed [11][12][13]. These coating techniques should ensure that the coated large-area perovskite thin films have the same quality as the spin-coated small-area ones regarding their microscopic morphology, crystal size, defect density, and vital optoelectronic properties, such as, carrier lifetime and effective diffusion length [14]. To achieve the same standards, there is still a need for vigorous encouragement of relevant research.
The scientific issues which play an essential role here are solvent volatilization, rheology, and fine regulation mechanism of key additives and coordination agents on perovskite nucleation, crystallization kinetics, and morphology evolution in the coating processes.

The inconsistency of high efficiency and high stability
Currently, the high-efficiency devices are those based on the regular n-i-p structure, of which nearly pure α-FAPbI 3 and doped Spiro-OMeTAD are used as the light absorber and HTM. The issue of phase instability in α-FAPbI 3 is welldocumented, in addition to which is the additional stability induced by doped Spiro-OMeTAD [19][20][21][22]. Two typical examples can be shown below. The first one was reported in Nature in 2021: a certified 25.2% power conversion efficiency (PCE) was achieved by the small solar cells based on the regular structure, as illustrated in figure 3. Even though high efficiency has been achieved, the stability is unsatisfactory. After 500 h of continuous illumination, the encapsulated devices preserved only about 80% of their initial PCE [23], which is too short for commercial application. The second example was reported in Science in 2021: a certified high efficiency of 19.3% was obtained by the regular structured module, but there is also the issue of poor stability (figure 4) [24]. The efficiency of the device soon droped to less than 80% of the initial value when the device was aged in dry air at 85 • C, which cannot also meet the requirement of IEC61215 stability standard. That's why it is crucial to find a suitable alternative HTM for Spiro-OMeTAD in order to simultaneously achieve high efficiency and stability in regular structured devices. The alternative direction is to further improve the efficiency of inverted p-i-n structured PSCs, of which the stability normally performs better (mainly because the charge transport materials of NiO x and C60 are dopant-free), but the highest certified efficiency (24%-25%) lags a little behind the regular ones. Furthermore, it is important to differentiate the stability of the α-FAPbI 3 main phase perovskite and whether or not it may attain similar photothermal stability to the existing FACsPbIBr-based perovskites with a little wider band gap and greater photothermal stability depending on various modification methods. In addition to the lack of stability of the material itself, the instability of the external environment will also worsen the device. It is very important to strengthen the further research of encapsulation technology to prevent moisture and oxygen in the external environment from seeping into the device and inhibit the volatilization of perovskite organic components, which is very important to improve the stability of the device [25].

The inconsistency among individual criteria of stability
Some devices passed standard aging tests in specific indicators, such as the damp heat test (85 • C/85% relative humidity for 1000 h) and thermal cycles test (−40 • C to 85 • C,   A 500 h stability test at 85 • C in ambient air (relative humidity ∼15 ± 5%) for the unencapsulated devices. From [24]. Reprinted with permission from AAAS. From [1]. Reprinted with permission from AAAS.
(figure 5) was used for stability tests, and the tested results showed that the device had excellent damp heat and humidity freezing cycle stability (figure 6(a)), which have passed the IEC61215 standards. However, the same devices showed poor light-soaking stability. The performance degradation occurred at a temperature as low as 35 • C under light soaking (figure 6(b)) [1]. The results clearly reflect that the impacts of thermal stressor and light stressor are different to this kind PSCs. Another typical example can be seen in figure 7(a). The device has a relatively stable p-i-n structure with the configuration of 'glass/ITO/poly[bis(4-phenyl)-(2,4,6-trimethylphenyl)amine]PTAA/perovskite/phenyl-C61butyric acid methyl ester (PCBM)/ZnO nanoparticles/aluminumdoped zinc oxide (AZO)/Ni-Al grid/MgF 2 ' [26]. The rear electrode made of sputtered AZO dense film also benefits the device's long-term stability because of its anti-corrosion property and non-permeable morphology. Devices with this configuration usually present a very long operation lifetime during light soaking under the normal temperature range (<50 • C). Therefore, one could find that after 1000 h of continuous illumination at 60 • C in an N 2 environment, the device's initial efficiency (13.6%) just declined slightly to 12.7%. In contrast, after an additional 320 h of illumination at 80 • C, the same device's efficiency degraded dramatically to 4.1%, and it increased to 4.7% after being stored in a vacuum in the dark for 10 months ( figure 7(b)). When comparing figures 7(c)-(e) to figure 7(b), it is clear that the device degraded significantly at 80 • C compared to 60 • C and that a huge area of the inactive yellow region appeared in the high-temperature light soaking device, while the device aged under normal temperature illumination at 60 • C hardly altered. This demonstrates that the device's operation stability varies significantly across temperatures (60 • C and 80 • C) and that the combined higher temperature and light soaking accelerates the degradation of the perovskite material. The stability test standards for crystalline silicon solar cells, as specified by IEC61215, have not considered the combined light and elevated temperature-induced degradation (LETID). The light soaking test in IEC61215 is also insufficient for the lifetime evaluation of PSCs, because IEC61215 just contains one exposure test corresponding to a luminous flux of 60 kWh m −2 , which is equivalent to 1 sun continuous illumination for only 60 h. The inherent reason is that the crystalline silicon is not easily degraded under light and mild heat from the natural environment, as the crystallization temperature of the silicon wafer (>1400 • C) and the doping temperature for the P-N junction (850 • C-900 • C) are Reproduced from [26] with permission from the Royal Society of Chemistry. much higher than the environmental value. However, organometal halide perovskites are particularly susceptible to photothermal stressors because the average crystallization temperature of perovskite films is only 100 • C-150 • C, resulting in its weak bonding energy and low activation energy for ion migration. The thermal stability of all-inorganic perovskite is significantly improved than that of organic-inorganic hybrid perovskite [27]. However, at present, the all-inorganic band gap deviates from the ideal value a lot. Pure CsPbI3 still has phase stability problems, and the efficiency is relatively low (the highest efficiency is 21.35%) [28]. In addition, allinorganic perovskite is much more sensitive to water, which requires higher encapsulation difficulty and higher humidity control requirements during manufacturing [29]. Furthermore, degradation of perovskites and devices may not be linear, as seen in figure 8, and may exhibit a leap when materials' damage accumulates to a certain extent. The trend of decline in the stability of different devices is unpredictable, and it is not appropriate to do epitaxial speculation of excessively long T80 lifetime from the initial 1000 h' aging test. Therefore, to prove the reliable photothermal tolerance of solar cells based on light and heat-sensitive light absorbers (such as perovskites in PSCs, organic semiconductors in organic solar cells, or dye molecules in dye-sensitized solar cells), sufficient light flux must be applied for light soaking tests, in addition to which, a specific high temperature (75 • C-85 • C) should be simultaneously considered [30]. The later raises a specific LETID test, which is of significant importance for PSCs since the land temperature in the Gobi desert can exceed 80 • C in the summer. Up to now, data on the ultra-long (10 000 h level) light soaking test and data on the LETID test of PSCs are still insufficient.

The contradiction between high performance and low cost
Many high-performance devices reported recently in the literatures, especially for high stability devices, still use Au as electrodes [32][33][34]. Au is relatively stable and reasonable to be used in the fundamental studies. However, it has no chance for such a noble material to be applied for mass production. Additionally, some organic interface materials are costly due to scale-up limitations or challenging synthesis and purification processes, making them impossible to lower their costs to an acceptable level. There is obviously no economic advantage in mass manufacture of such kind materials. According to some commercial companies in China, perovskite thin films in PSCs can be produced at around 10 yuan m −2 . Considering 2 g perovskite (500 nm thickness) is required for 1 m 2 module fabrication, 10 yuan m −2 implies the unit price of perovskite has to be lowered down to 5 yuan g −1 . Chinese chemical suppliers could reach this price quickly since the standard synthesis and purification procedures for PbI 2 and formamidinium iodide (FAI) are pretty simple. There are already some manufacturers of perovskite materials which could sell their highpurity PbI 2 and FAI products at the prices of <10 yuan g −1 .
Similarly, low prices could also be expected for inorganic interface materials, such as NiO x or SnO 2 . In high-efficiency PSCs right now, some organic interface materials with higher  unit costs are unavoidably used. It is important to think carefully about how to approach their economic situation. The current cost of a kilogram of the raw material for C60 in high-efficiency inverted PSCs is 150 yuan g −1 , much lower than its derivate PCBM. The price of PCBM is hard to be lowered due to complex synthesis and purification procedures, while that of C60 could be quickly lowered due to scale production. The cost of the evaporated C60 film will be at least 45 yuan m −2 if the maximum material utilization rate is 30% and the C60 thickness is calculated at 40 nm (equal to 0.3 g m −2 of C60 will be used). This price should be primarily lowered in the future if compared with perovskite and other interfacial materials. Other organic interface materials, such as some self-assemble monolayer (SAM) or PTAA, typically have a thickness of 5-40 nm. To be more cost-effective, the costs of these organic materials should be kept to a range of 150-300 yuan g −1 (even lower is better), corresponding to a price of 5-30 yuan m −2 , depending on the material utilization rate of coating or printing. Therefore, the pricing must be carefully considered in order to determine whether it will provide a competitive advantage after mass production in the development of efficient and stable organic semiconductor interface materials. Another important factor that may affect the cost reduction rate of PSCs is the current high price of TCO glass. The price of commercial FTO glass (standard quality, >80 yuan m −2 ) from the Chinese glass market is still slightly higher than the PSC producer's expectation. This price has the chance to be quickly lowered to 40-50 yuan m −2 . At last, the automatic production line of PSMs with high production efficiency and a controlled rate of defective products is also essential to lower the module price. Based on conversations with materials and equipments manufacturers, a fast cost reduction rate of 100 MW PSCs production line could be expected. The prices of active films in PSMs could be expected to be 10 yuan m −2 (for the perovskite layer) plus 25 yuan m −2 (for interfacial layers) plus 40 yuan m −2 (for FTO glass) plus 10 yuan m −2 (for rear electrode), which is 85 yuan m −2 in total. The encapsulation materials, including hot melt adhesive membranes, encapsulated cover glass, etc, should be similar to or a little higher than that of crystalline silicon solar modules, which could be expected to be about 50 yuan m −2 . If further considering the equipment depreciation, labor, and energy costs of 45 yuan m −2 in total, the over-all module price could reach 85 + 50 + 45 = 180 yuan m −2 . In addition to evaluating the price of key materials in the Chinese market, we have also carried out corresponding statistical analysis on the cost of materials in other regions in the world, as shown in table 1. If a mass production module efficiency of 18% (180 W m −2 ) could be realized, the module price of PSMs could quickly reach 1.0 yuan W −1 , which is ∼1/2 of the current module price of crystalline silicon solar modules. In such a scenario, the cost advantage of PSMs will be determined first among the 'golden triangle' parameters. Among many energy technologies, the power generation cost of solar cells is relatively cheap. Green analyzed the reasons. The research shows that the improvement of component efficiency is the primary factor of cost reduction, followed by the reduction of material consumption, which further reduces the cost of the device. In addition, in the context of specific events, the government's macro-control and financial support have also promoted the further development of photovoltaic power generation technology and reduced the cost of solar cell power generation [35].

Conclusions and future perspectives
In short, for the commercial application of PSCs, the main bottlenecks rest with two aspects: (a) large-area PSM's high efficiency; (b) competitive long-term stability. It deserves discussion on whether it is necessary for the low-cost PSMs to reach the same lifetime of 20-25 years to the mature crystalline silicon solar modules in the initial stage of PSMs' commercialization. PSMs may establish its own sales model and business logic because they could be really cheaper than other types of solar cells. And the next big things are to resolve the four distinct contradictions: (a) the overlarge efficiency gap between small-area and large-area PSCs; (b) the inconsistency of high efficiency and high stability; (c) the inconsistency among individual criteria of stability; (d) the contradiction on high performance and low cost. All of these are closely related to the levelized cost of efficiency (LCOE) of PSMs. Only when LCOE of PSMs could compete with other mature photovoltaic technologies, PSMs could truly realize its commercialization.
In view of these contradictions in PSCs, we propose to solve them mainly from the following aspects: (a) the uniformity of large-area perovskite films is poor and there are many defects, resulting in the PCE of large-area devices is lower than smallarea. It is important to strengthen the research on the crystallization kinetics of thin films and control the grain growth process by means of additive engineering and anti-solvent engineering. (b) The instability of high efficiency solar cells based on Spiro-OMeTAD (HTM) is mainly due to the unstable dopant in them, which is easy to absorb water or evaporate, and even react with perovskite to accelerate the degradation of devices. Looking for additives with more stable performance or suitable HTM replacement materials will help to improve the stability of the device. At the same time, we can further improve the efficiency of inverted p-i-n structured PSCs with more stable performance. Strengthening the research of encapsulation technology to improve the stability is also important. (c) The test index of PSCs is determined based on the IEC61215 standard of crystalline silicon solar cells, which is not comprehensive and perfect for PSCs. The establishment of a test standard belonging to PSCs is crucial to promote their commercial development. Recycle expensive and chemically stable Au and TCO or find cheaper electrode materials to replace Au, such as Al, Cr and Cu metal electrodes, so as to reduce the manufacturing cost of devices. In addition, try to improve the stability and PCE of devices to reduce the LCOE of PSMs.
Among these challenging issues, it is thought that the stability issues are more critical and more urgent to be resolved, rather not large-area high efficiency, because the theoretic uplimit of PSC's stability has some uncertainty; even in small area research cells in labs may not enough to support a long enough outdoor lifetime of PSMs. One has to challenge light soaking test under 1 sun for up to 10 000 h which equals to luminous flux of ∼10 years for real outdoor test. LETID test at 75 • C-85 • C is also required to pass for PSMs. Potential induced degradation at ±1000 V and hot spot effect in PSMs, which have not been intensively studied yet, should be strictly evaluated. And more important is to conduct sufficient outdoor test in different climates conditions over years. Only after completing these, the real operation lifetime for PSMs in a specific application scenario could be clearer. For large-area high efficiency, though it is also very challenging, it is thought it is all about time since small lab cells' efficiency has been demonstrated high enough. In order to resolve this, to develop new solution inks and high precision machines for large-area perovskite film coating are essential. Only if the same quality of large-area perovskite films could reach the same level of spin-coated small-area ones, the other challenges on other functional layers' uniformity and laser etching's controllability are not so difficult.
Besides cost-stability-efficiency, lead toxicity in lead halide-based PSCs/PSMs is another potential obstacle for the commercialization and large-scale applications of this emerging PV technology [40]. However, there are contradictory opinions about lead toxicity in lead halide-based PSCs. On one hand, Pb release from PSCs is harmful due to its nonbiodegradable and highly bioavailable nature [40]. It is thus important to replace lead with low-toxic materials [41,42] or use lead trapping/blocking materials [43,44] to reduce the toxic risk of PSCs by in PSCs. However, the former strategy usually leads to lower efficiency when compared with its pure lead counterpart and the latter one will further increase the cost of the PSCs. On the other hand, another opinion says that the released lead amount from PSCs actually is very low because of the low usage amount of lead, which is less than 1 g Pb/1m 2 . Even if the lead leached out from PSCs, the lead will spread, making the concentration of lead into the environment to be less than that of a coal-generation station [40,45]. Thus, further fulllifecycle evaluation of the impact of the lead toxicity issue in PSCs and strategies to further reduce such issue is needed to guarantee the successful commercialization of this emerging PV technology.