Meniscus fabrication of halide perovskite thin films at high throughput for large area and low-cost solar panels

Halide perovskites have rapidly attracted considerable attention due to unprecedented properties not seen in traditional semiconductors. In addition to their optoelectronic merits, one advantage of perovskite materials is their solution processability, which opens the door to low-cost and high throughput solution coating strategies for the commercialization of perovskite solar cells (PSCs). Here we review perovskite film fabrication by meniscus coating—a simple and readily scalable manufacturing technique, including blade coating and slot-die coating. We outline the fundamental fluid mechanisms of meniscus coating, discuss drying and crystallization of perovskite in the coating process, and provide an overview of recent progress in meniscus-coated PSCs.


Introduction
As the technology rapidly advances and the costs reduce, the demand for solar panels continues to grow. Nearly 100 GW of grid-connected solar panels were installed in 2017, an almost 30% increase over the 77 GW added in 2016 [1]. Analysts predict that the world's photovoltaic (PV) production capacity could reach one terawatt by 2022 [1]. At the same time, various new markets are emerging for cost-efficient renewable power, such as portable power supplies, electric vehicles, and building-integrated PVs. The everdecreasing price of solar energy plays an important role in promoting clean energy in these rapidly developing markets. The projected total system levelized cost of electricity (LCOE) of solar power is currently $48.8/MWh, according to the US Energy Information Administration [2]. The estimated cost dropped 88% from 2010 to 2019. Many developments in the industry have contributed to that price reduction, but now it is new thin film PV technologies which are emerging with great promise to further reduce the cost of solar energy.
Halide perovskite solar cells (PSCs) is one such emerging PV technology. They have received considerable attention in recent years due to perovskite's intriguing optoelectronic properties including long charge diffusion length, high carrier mobility, and large absorption coefficient [3][4][5][6][7]. Because of these qualities-among others-the power conversion efficiency (PCE) of small, single-junction PSCs has rocketed to 24.2%, which is already the highest of all thin film PV technologies [8]. The high optical absorption coefficient of up to 10 4 cm −1 is a critical merit for commercialization; films of just 300-1000 nm are sufficient to collect the photons sufficient for achieving high efficiencies. Compared to inorganic thin film absorber layer thicknesses of around 3 μm, it not only reduces material use and processing costs but also enables high production throughput. Han et al calculated the manufacturing cost of perovskite modules [9], with estimates as low as $30/m 2 to $41/m 2 without considering panel components such as sealing the glass and junction box. The corresponding LCOE is quite low, only $35-$49/MWh with a module efficiency between 12% to 20% and a 15 year lifetime. However, to value the commercial product, the cost needs to consider the total panel system. Heben et al established a more comprehensive model which considered assembling the panel components. By using a printing strategy to fabricate the hole transport and electron transport layers (HTL and ETL, respectively), the production throughput can be 1.44 m 2 min −1 . Thus, the annual capacity could reach 121 MW year −1 with a module efficiency of 16%. The LCOE is in the range of $49.3 to $79.0/MWh assuming a system lifetime of 30 years [10].
To scale up and reduce manufacturing costs, effective mass production methods should be implemented to fabricate perovskite films. Several versatile film deposition methods have been reported to fabricate high-quality perovskite films for PV applications; these can be classified as a vapor deposition method (typically conducted by thermal evaporation) or a solution-based method in terms of the phase of precursor material used for manufacturing. For example, coevaporation of PbI 2 and CH 3 NH 3 I (MAI) yields a uniform, compact perovskite layer on the substrate. However, this mode of film deposition faces significant challenges that limit the scale of film production. Tarasov and collaborators innovated a new vapor evaporation strategy by taking advantage of the redox reaction of reactive polyiodide melts and lead to form perovskite [11]. Thin film Pb and MAI were first deposited stoichiometrically on a substrate by thermal evaporation. The as-deposited film is then treated with I 2 vapor, converting the Pb/MAI bilayers into MAPbI 3 . Devices fabricated by such strategy achieved PCE of 17.18% with the capability of making large films up to 600 cm 2 . Another feature of the vapor method is that it can form a conformal thin film on textured substrates. Oxford PV exploits this merit to fabricate perovskite/silicon tandem solar cells, achieving 28% record PCE in 1 cm 2 (table 1) [12]. Although vapor deposition can form relatively large area films, it is not quite compatible with mass production due to sophisticated vacuum equipment and long processing times, reducing the final throughput. Due to these limitations, most of the institutes and companies are focusing on solution strategies to commercialize PSCs. Solution methods are well known, low-cost strategies that are capable of processing at low temperature under ambient conditions. Spincoating is the widest-used solution method; almost all record efficiencies for PSCs are based on this method. However, spincoating has very low throughput and wastes a large amount of precursor solution-not suitable for mass production. In contrast, other solution techniques like blade coating [13][14][15][16], slotdie coating [17][18][19][20], spray coating [21][22][23], and inkjet printing [24][25][26] are viable options for the scalable fabrication of perovskites. In industry, Solaronix and Saule Technologies have made efforts (shown in table 1) on printing strategies, which allow printing in custom patterns. Solaronix has achieved 12% PCE on modules with an area of 500 cm 2 [27]. However, whether they are suitable for large area mass production will depend on the printing speed and device structure [28]. Other companies have implemented meniscus coating, a simple coating method readily compatible with high throughput production. The term meniscus refers to the formation of the meniscus shape during coating of the solution. The meniscus configuration and fluid flows within it critically influence the morphology of the final solid film. Common meniscus coating methods include dip coating, blade coating, and slot-die coating. The Dutch firm Solliance demonstrated their roll-to-roll (R2R) slot-die coating capabilities, yielding large-scale perovskite production; their technique produced solar cells with a PCE as high as 13.5% for a device area of 0.09 cm 2 [19]. Toshiba produced the world's largest perovskite module: 702 cm 2 -with a module efficiency of 11.7% [29]. The highest certified efficiency for a perovskite module in industry is a 17.25% PCE mini-module (17.3 cm 2 ) manufactured by Microquanta Semiconductor in China [30]. However, they did not report the film fabrication technique.
In this review, we summarize recent works implementing meniscus coating of perovskite films, because it is an efficient, low cost and high throughput film deposition technique that has been well established in many industries, though still relatively new for perovskites. An overview of the meniscus coating mechanism is described. The evolution of the perovskite from the initial wet meniscus to the solid, crystalline film is analyzed step-by-step. Works related to each phenomenon will be reviewed. Based on the fundamental understanding of the meniscus coating process, we provide perspectives on the development of the large area and low cost perovskite solar panels.  [32]. On the other hand, planar n-i-p and p-i-n structure without the mesoporous layer have proven to be efficient architectures as well. The highest reported efficiency of planar PSCs is 23.3%, with the aid surface passivation [33]. The planar architecture has excellent potential for low-cost mass production, due to its simple structure and circumvention of a high-temperature annealing process for the mesoporous metal oxide.

Meniscus coating strategies
Meniscus coating can be defined as the translation of a meniscus over the surface of a substrate as the solution is spread by the coating tool. The shape of the meniscus, solution flows within it, and the solvent drying speed during and after coating the meniscus dramatically impact the morphology of the final, solid film. Common meniscus coating methods include dip coating, blade coating, and slot-die coating. Blade coating and slot-die coating are the most promising and reproducible strategies for the scalable deposition of perovskite films; they are also amenable to rollto-roll processing, which possesses great potential for manufacturing large area solar panels at high throughput for mass production.
2.2.1. Dip coating. Dip coating is a process that first immerses the substrate in the precursor solution and then lifts it out from the solution vertically with a specific velocity (figure 2(a)). A meniscus forms on the surface of the substrate at the interface between the precursor solution reservoir and the air (or inert atmosphere). Despite its simple appearance, complex interactions influence the film's structure and thickness, such as withdrawal speed, functionality of substrate surface, gravity, surface tension of solution, environmental temperature and humidity, as well as other factors [35]. In addition, the crystallization and phase transition of the perovskite will make the process far removed from handling. Also, the batch-to-batch immersion is a slow process that would hinder mass production. These drawbacks make dip coating an unpopular method for perovskite film fabrication.

Blade coating.
Blade coating is one of the most versatile ways to spread the solution over a substrate. It is also known as 'doctor blading' or 'knife coating', because a doctor blade or knife is the primary instrument involved in the spreading process [36]. The coating solution is initially applied in the gap between a coating head and a substrate. Next, the blade travels across the substrate, spreading a uniform wet film and removing excess solution, as illustrated in figure 2(b). The wet film thickness can be related to the gap distance and also to the coating speed. Menisci form at both the advancing and receding sides of the blade. The advancing meniscus impacts the spread of solution onto the substrate which can limit the coating speed. Within the receding meniscus is a complex combination of different solution flows; this area largely determines film thickness and uniformity.
2.2.3. Slot-die coating. In slot-die coating, the coating solution is forced through a coating die onto the substrate as showing in figure 2(c). The coating die head, as well as the distributor in the system, consists of a distribution chamber and a feed slot which provides a uniform liquid flow rate over the entire coating width. The solution viscosity and capillary force dominant the film spreading. The coating gap underneath the downstream die is filled by solution whereas the upstream coating gap can be partially or completely filled [37]. Menisci form at both the upstream and downstream sides as in blade coating.

Advancing meniscus
The advancing, wetting meniscus is typically considered less than the receding meniscus since the latter is where the wet film forms and drying and solidification occur. However, the advancing meniscus is still valuable to study as it is the location of the first contact between the coating solution and substrate, and this wetting dynamic plays a vital role in limiting the coating speed.
Wetting is an essential process of coating-initially, coating head and substrate contact air. The precursor solution is injected, filling the room between the head and substrate. This spreading process is called static wetting, which also affects the coating process, though less directly. During coating, the liquid displaces air from the moving solid surface. The advancing meniscus obeys the rules of dynamic wetting. A straight and steady wetting line is crucial for uniform coating. Many theories have been established to describe static and dynamic wetting [38][39][40].
For static wetting under ideal conditions, smooth and chemically homogenous solids, and pure liquids, the thermodynamic equilibrium of gas, liquid and solid phases determine the unique static contact angle θ at which the liquid meniscus intersects the solid surface (figure 3(a)). According to Young's equation [41], the relationship between the contact angle θ, the solid-gas interfacial tension σ sg , the solidliquid interfacial tension σ sl , and the liquid-gas interfacial tension σ lg is: If the vector sum of surface tension σ sg , σ sl , and σ lg on the solid surface has the same direction as the solid-gas tension, then the liquid can spread on the solid spontaneously to form a film. The Wenzel [42] and Cassie [43] equations, which are the modification of Young's equation for heterogeneous substrates, are also well used to solve the wettability of the solid surface.
In the coating process, a wetting line moves with respect to the solid surface. The dynamic pressure and surface tension modulate the advancing meniscus of the wetting line. Dynamic wetting is usually characterized by the relative velocity of the wetting line and the dynamic contact angle between the dynamic liquid line and the solid surface. Dynamic wetting includes both spontaneous wetting and forced wetting. Forced wetting is the basic situation in meniscus coating, in which the contact line is driven by the external force of the coating head. Generally, the dynamic contact angle is velocity-dependent, which differs from its spontaneous wetting counterpart. As a consequence, dynamic wetting failure can be caused by an excessively high coating speed. The solution fails to form a continuous wet layer when the coating speed exceeds the critical speed U crit , resulting in unsteady flows and air bubbles within the liquid [44]. Thus, manipulating dynamic wetting is one of the keys to highspeed coating. However, few studies have intensively investigated the high-speed coating of semiconductor materials, particularly perovskites.

Receding meniscus
Compared to the advancing meniscus, receding meniscus dynamics significantly influence the coated film quality. Various flows in the receding meniscus have been well investigated by researchers [45][46][47]. In the coating process, the coating head produces a shear force on the precursor solution filled in the space between head and substrate due to the relative movement. This shear force induces a shear-driven fluid motion called Couette flow [48] ( figure 3(b)). The flow direction is parallel to the coating direction. Moreover, when considering the pressure gradient between the advancing and receding meniscus, the pressure differential can generate another force overlapping the shear force, tuning the flow magnitude and direction. A favorable pressure gradient in the direction of motion enhances the Couette flow. On the other hand, an adverse pressure gradient can weaken the Couette flow, even causing a reverse flow at the area far from the moving blade, where the shear force is weak.
Another primary flow is the Marangoni flow [49] triggered by the surface tension gradient along with the meniscus liquid-air interface (figure 3(c)). As the solvent evaporates, a higher concentration of solute near the solid film increases the surface tension, causing Marangoni flow toward the dry edge. Conversely, a reverse direction Marangoni flow can be induced by the thermal gradient from the dry edge to the bulk solution when heating the substrate during coating. Such a temperature gradient induced Marangoni flow can be characterized by the Marangoni number, a dimensionless number comparing the surface tension and viscous forces [50] g ma where γ is surface tension, T is temperature, L is effective length between the bar and substrate, μ is dynamic viscosity, and α is thermal diffusivity. Moreover, a faster evaporation rate close to the liquid-solid contact line induces the 'coffee ring' effect [51,52]. A capillary flow can be created to equilibrate the fast solution losing at the edge. Surface tension induced Marangoni flow is negligible in a single solvent system. However, solvent engineering of perovskite precursor solution usually involves mixed solvents, which can introduce surface tension gradients during evaporation, inducing Marangoni flow. Marangoni flow has been used to describe film formation in the spray coating of perovskite layers. It was regarded as the main mechanism of self-organization of gaps between spayed droplets. Tung et al [53] claimed the volatile difference of dimethyl sulfoxide (DMSO) and N-Methyl-2-Pyrrolidone (NMP) in solution leads to surface tension gradient from the edge to the center of each droplet. Therefore, driving force forms and spreads the droplets to mobilize the initially static pinning lines and recover the unwetted areas and pinholes, yielding compact films.
In addition, surfactants can also impact surface tension. When surfactant exists in solution, the energetic dynamics force the surfactant to reside at the free surface. As a result, the presence of surfactant at the surface reduces the surface energy. This surface tension gradient, in turn, causes Marangoni flow. We found such a mechanism can be used in blade coating perovskite to obtain large scale, uniform films [14]. In situ observation of the solution drying process immediately after blade coating shows that the intermediate phase solute moves to a crystalline perovskite island in 5 s. The solute flows toward the center of islands, leaving gaps between each island. To form compact perovskite layers, surfactants were introduced to the precursor solution. Consequently, the transportation of solute also carries the surfactants to the central island, creating a surfactant concentration gradient. The high concentration of surfactant in center reduced the surface tension, generating Marangoni flow in a direction reverse to original solute flow. Subsequently, the equilibrium of the two flows suppresses gaps between islands.
Meniscus flows can also induce alignment of coating materials when deposition speed falls in evaporation regime. Details will be described in the next section.

Coating regimes
Depending on the variation trend of solid film thickness, two coating speed regimes are identified in experimental results. In the low-speed regime, namely the evaporation regime, the drying timescale is comparable to moving speed. The solute accumulates and forms solid film right in the meniscus area ( figure 4(a)). On the contrary, Landau-Levich regime is where coating speed is high enough so that wet films form and remain on the substrate due to the relatively low drying speed compared to the coating speed [54] ( figure 4(b)).
These two deposition regimes were first combined to explain the thickness variation of blade coated phospholipid films [55]. The same phenomenon has been found in the meniscus coating of perovskite [14], as shown in figure 4(c). When the coating speed is low, the evaporation regime governs the coating process. The dry film thickness follows a power-law u µt , 1 where the solid particles arrange right below the meniscus. The film thickness decreases as increasing the coating speed due to shorter residence time per unit length, leading to a reduced amount of dried material. Drying at the meniscus position leads to fluid flows in the meniscus which strongly impact film formation. For example, Lin et al [56] invented a meniscus-assisted solution printing (MASP) strategy to yield large grain, dense perovskite films. The extremely slow coating speed of 12 μm s −1 place their coating process within the evaporation regime ( figure 5(a)). Simultaneously, fast solvent evaporation rate at the edge of the meniscus triggered 'coffee ring' effect, transporting perovskite solutes to the contact line, facilitating the growth of micrometer scale grains ( figure 5(b)). The solar cells fabricated by this strategy achieved a PCE approaching 20%, benefiting from the low trap density and long carrier lifetimes in the large grain perovskite film. Also, the MASP coated film shows a preferred crystal orientation since the meniscus flows heavily influence the solid film formation at such a low coating speed.
The uniaxially oriented film fabricated by meniscus coating has also been reported in perovskite coating. Van Hest et al [57] claimed preferred orientation of the blade coated film at a speed of 2.5-3 m min −1 . For solution dissolved by N,Ndimethylformamide (DMF), the coating speed was relatively high, so that evaporation at meniscus area was not fast enough to wholly dry the solvent. However, the authors invented a new solvent by methylamine and acetonitrile, with a very low boiling point and high vapor pressure ( figure 5(c)). The fast evaporation of solvent induces an evaporation regime coating process. The Couette flow due to shear force, along with Marangoni and capillary flows drive the growth direction of perovskite film. Consequently, the blade coated film has highly oriented 2D XRD patterns (figure 5(d)) compare to spin-coated film, which consistent with strong periodic XRD peaks that correspond to the (hk0) of MAPbI 3 .
However, not all the reported perovskite films fabricated by the meniscus coating method have a uniaxial orientation. The reason why some reports demonstrate highly oriented films while others do not is due to the coating regime the process employs. Liu et al [58] specifically investigated the relationship between meniscus coating speed and film orientation of perovskite films. Results proved that films coated at slow speeds, such as 0.2 mm s −1 , exhibit highly aligned morphology. The evaporation within the meniscus region at such slow speed dominates film formation, inducing flow governed orientation. On the other hand, raising speed tremendously changes the solid film morphology, forming large isotropic plate domains. Figures 5(e) and (f) show the apparent morphology difference at various coating speeds. GIWAX data of slow speed coated films confirmed predominant reflection plane is (110) which is parallel to the substrate. Obviously, the preferential orientation originates from the confined nucleation and growth of the moving blade.
Nevertheless, coating film in the Landau-Levich regime is the goal for real production, as faster fabrication is optimal for mass production. Contrary to the slow speed, the faster coating speed of the Landau-Levich regime has a trend of increasing film thickness with coating speed. Wet film forms on the substrate since the coating speed is higher than the solvent drying rate. The separation of wet film spreading and solidification merits the formation of a uniform solid film. However, dewetting of the as-casted wet film before drying can also ruin the solid film uniformity.

Fate of the wet films after spreading
Unlike the spin-coating process, in which the solvent casting, thinning, and drying all happen at the same place, the meniscus coating in the Landau-Levich regime has spatially separated casting, leveling, and drying processes. The casting process has been comprehensively discussed above. Nevertheless, the fate of the wet layer after casting determines the crystalline film quality. The evaporation of the solvent, nucleation, and growth of solid film play critical roles in solid film formation.
3.4.1. Dewetting. The coating processing of the Landau-Levich regime benefits the healing of irregularities of film thickness originated from the meniscus area. The interval before wet film drying gives time for the liquid film leveling under gravity. However, a spread liquid film can also undergo a dewetting process which ruptures the wet film. Local dewetting can be caused by evaporation, capillary flow, or surface tension gradients. Once the wetting film is sufficiently thin, long-range intermolecular forces can expedite the thinning until cracking occurs. Further, during film solidification, the ruptures can be DMSO is a commonly used solvent for perovskite precursor solution because it can readily coordinate with lead halide (e.g. PbI 2 ). The intermediate phase formed by this strong interaction retards the crystallization which encourages the formation of a uniform and dense perovskite film [59]. However, DMSO has high surface tension and a high boiling point of 189°C. This leads the wet film to contract under surface tension before drying, resulting in holes and gaps in the solid film. To solve this problem, Galagan et al [19] mix cosolvent 2-butoxyethanol (2BE) with DMSO to R2R slot-die coat methylammonium lead iodide and other mixed cation and mixed halide perovskites. 2BE has a low boiling point and small surface tension. It not only reduces the net surface tension, improving the wettability of the wet layer but also increases the evaporation rate, leading to fast drying and mitigating the dewetting phenomenon. Besides, the authors further accelerate solvent evaporation rate by setting a faster  temperature ramp-up after coating. Through these strategies, the solidification of the wet layer has been well controlled. A compact plate-like perovskite film fabricated by R2R slot-die coating on ITO-PET achieved highest PCE of 13.5%.
We demonstrated that surfactant is another strategy to avoid dewetting of the blade-coated wet film [14]. Usually the hydrophobic properties of hole transportation material poly (bis(4-phenyl)) (2,4,6-trimethylphenyl) amine (PTAA) make it challenging to form compact perovskite layer. However, the surfactants in perovskite solution effectively modified the ink/substrate interface and improved the affinity of perovskite solution to substrates. In this way, the surfactant facilitates the formation of fully coverage perovskite layer on the hydrophobic PTAA by blade coating.
After forming a uniform, fully covered wetting layer, several phenomena affect the evolution of the film, including solvent evaporation, thinning of wetting layer, phase transition, and solidification. Each of the processes can critically influence the morphology and electrical properties of final crystalline perovskite film.

Drying and crystallization.
Although researchers have long known that controlling the solidification process is an essential key in achieving a high-quality perovskite film, it is only recently that researchers used in situ characterization techniques to elucidate the process in the transition from precursor solution to solid film while spin-coating or meniscus coating. In situ characterizations unveil that unlike polymer semiconductors, perovskite undergoes a unique phase transition during the solution coating process. The so-called intermediate phase which forms because of the coordination of solvent and lead halide plays a critical role in film formation.
In situ timeresolved grazing-incidence wideangle x-ray scattering (GIWAXS) is a powerful tool to investigate the phase behavior in the film coating process. Amassian et al [60] first observed the phase transition in spin-coating MAPbI 3 in DMF. A highly ordered intermediate phase out of the disordered precursor formed 20 s after casting, which shows three intense peaks at low scattering vector q values. The phase is stable without additional treatment. The morphology of such films exhibits a needle-like structure, which stems mainly from the preferred orientation of intermediate phase crystal. The ordered intermediate solvate phase and needle-like morphology have also been observed in films blade coated by the same ink formulation at room temperate [61]. However, heating the substrate during bladecoating has been found to significantly alter the phase transition process. Substrate temperature higher than 100°C can completely circumvent the solvate phase, which mainly due to the fast removal of the solvent in the as-casted wet film. GIWAX shows a direct transformation from disordered colloids to perovskite phase. A compact film with large domains rather than needle-like morphology has been formed. Not surprisingly, the PCE has improved because of the compact morphology by high-temperature blade coating. Recently, Liu et al [62] elucidated the identical key point for forming a high-quality perovskite film by spin-coating and blade-coating: suppressing intermediate phase over-crystallization, regardless of coating temperature. Intermediate phase is widely accepted as a pivotal to growing large perovskite crystals [63]. However, the over-crystallization of intermediate phase distorts the morphology to textured, incomplete coverage. As a consequence, the anti-solvent dripping method must be controlled precisely due to the narrow operation window of supersaturating solvent without solvate phase over-crystallization ( figure 7(a)). In the hot substrate scenario, the high temperature helps evaporate solvent immediately after coating. The extremely fast evaporation directly transits disordered sol-gel to perovskite phase, circumventing the intermediate phase ( figure 7(c)).
So far, several strategies have been established to realize controllable solvent removal rate, including anti-solvent, hot substrate, and gas quenching. The basic ideas are all based on suppressing over-crystallization of intermediate phase or even completely avoiding it.
3.4.2.1. Antisolvent. As mentioned above, anti-solvent dripping has been widely used in spin-coating to produce high-quality perovskite films. However, the common dripping process is challenging to implement with the meniscus coating. However, instead of dripping, Zhu et al incorporated blade coating with anti-solvent bathing method [64,65] (figure 8(a)). Unfortunately, the timescale of transferring the as-casted wet film to the solvent bath is much longer than the intermediate phase formation time. As a consequence, bad morphology appears even with the anti-solvent applied. On the other hand, additives such as methylammonium chloride (MACl) are commonly used to control perovskite morphology [66,67]. The authors take advantage of MACl to prolong the operation window for antisolvent treatment. An 8 min wet-film stage can be achieved by adding excess MACl to the MAPbI 3 precursor solution, which decouples the precursor film casting and perovskite formation, enabling the antisolvent extraction. The devices made by such strategy achieved PCE of 18.55% with 0.12 cm 2 area, and 13.3% active-area efficiency of a 12.6 cm 2 module.
3.4.2.2. Hot substrate. Heating substrate is a simple way to facilitate solvent evaporation ( figure 8(b)). It has been widely used ever since the early stages of study on meniscus coating perovskite [13-16, 68, 69]. Boiling points of commonly used solvents DMF, DMSO, GBL, NMP are 152°C-154°C, 189°C, 206°C and 202°C-204°C respectively. To remove these solvents completely and effectively, the substrate must be kept at a temperature close to the solvent's boiling point. Greater than half of all works reviewed used the hot substrate strategy to meniscus coat perovskite (table 2). The highest device efficiencies based on meniscus coated perovskite also implement heating of the substrate during coating. Recently, we incorporate bilateral alkylamine (BAA) as a multifunctional additive in perovskite precursor for blade coating with 150°C substrate temperature. The optimized device achieved a champion PCE of 21.7% with 21.5% stabilized output [70]. Moreover, we also showed the  capability of hot substrate blade coating strategy on making perovskite modules. By combining the surfactant additive as mentioned before, we demonstrated perovskite modules of up to 14.6% efficiency for a 57.2 cm 2 aperture area [14].
3.4.2.3. Gas quenching. The gas quenching method is believed to have good reproducibility as it avoids using chemicals and is a low-temperature process [71]. The flowing gas takes away the solvent from the as-coated wet layer on substrates, leaving perovskite materials to form the solid film ( figure 8(c)). Many excellent results based on this method were reported in spin coating techniques; meniscus coating can also take advantage of this strategy. Vak et al [20] slot-die coat perovskite film by two steps method. The N 2 gas was introduced to accelerate the solvent evaporation during slotdie coating the PbI 2 layer, forming a pinhole-free uniform film. After that, MAI was employed by slot-die system to convert PbI 2 to perovskite. A PCE of 10.14% was demonstrated, which is the highest efficiency of a fully slotdie coated PSC. Other works integrate the gas-quenching with a hot substrate to facilitate solvent evaporation. A heater was equipped on the slot-die machine to heat the substrate. A gas blower was housed behind the slot die-heat to flow N 2 above the film surface. Benefitting from the gas flowing, the substrate temperature must not be as high without the gas quenching. Temperatures ranging from 45°C to 65°C were found to be the optimized parameters [20,72,73]. The best PSCs fabricated by such method achieved PCE of 15.57% [72]. Moreover, gas quenching can be facilely employed without doctor blading. Hu et al [74] blew N 2 gas on a wet perovskite film directly ( figure 8(d)). The gas can blow away the solvent efficiently due to the low moving speed (5 mm s −1 ) of the gas knife. The devices made by fully air-blading method achieved PCE of 20.8% for 0.09 cm 2 and 19.12% for 1.0 cm 2 . While this review is preparing, Li et al combined the gas quenching and blade coating, achieving a hysteresis-free PSCs with PCE of 20.26% for 0.06 cm 2 [75]. The air knife was installed in a way that laminar nitrogen blows in a horizontal direction. This setup not only provides reproducible processes but also enables in situ measurements on the film's drying and crystallization.

Meniscus coating mix cation/anion perovskite
Up to now, most devices fabricated by the meniscus coating method still work on MAPbI 3 , as a single composition is easy to control. However, mixed cation and anion perovskites have been well known to have numerous advantages that single component perovskites cannot reach, including higher performance, increased stability, enabling band gap tuning, enhanced charge carrier transport, etc [76]. Formamidinium (FA) in MAPbI 3 can reduce the bandgap, enabling the devices to approach the optimization bandgap of Shockley-Queisier limit [77]. However, FA perovskite is thermodynamically stable in a δ phase at room temperature, which needs enough energy to overcome the phase transition barrier to perovskite phase. Inspired by the hot substrate blade coating, which enables heating the substrate and blading precursor solution simultaneously [15], we successfully made FA and MA mixed perovskite by the blade coating method. The high-temperature substrate not only helps accelerate solvent evaporation but also overcome the energy barrier of δ phase to α phase. Mixed cation perovskite FA 0.4 MA 0.6 PbI 3 reached a bandgap of 1.55 eV. The champion cell yielded a PCE of 18.3%, which is higher than the MAPbI 3 devices fabricated by the same process. Cesium is another widely used A site element. Cs ion was designed to substitute part of organic ions to improve thermal and structural stability [78]. Fortunately, Cs ions in combination with MA were found to facilitate crystallization of precursor solution to solid perovskite phase at room temperature [79]. As a result, it is possible to fabricate triple cation perovskite by the meniscus coating method without heating the substrate. Furthermore, Cs interacts strongly with [PbI 6 ] 4− which replaces the intercalated DMSO, reducing the concentration of intermediate phase and promoting rapid crystallinity.
At the halide site, chloride was first introduced in MAPbI 3 to obtain outstanding PV performance. However, comprehensive studies revealed that the big halogen ionic radii difference hindered the incorporation of Cl from forming perovskite phase [80]. Instead, the real role of the Cl ions is as a dopant that influences the crystallization process. This property has been used in meniscus coating as mentioned previously [64]. Contrary to chloride, bromide can alloy with iodide, increasing perovskite bandgap up to 2.3 eV [81]. We also demonstrated that incorporating Cs and Br in perovskite precursor solution can reduce the required film formation temperature for blade coating [82]. A blading temperature of 120°C was used to form a high-quality film of MA 0.6 FA 0.38 Cs 0.02 PbI 2.975 Br 0.025 perovskite, compared to 150°C for MAPbI 3 composition. Such mix cation and halide film achieved stabilized PCE of 19.3% with less than 10% degradation after storage for one month.
Although mixed tin-lead perovskite is another hot topic in perovskite research-due to potential in tandem solar cell applications-few reports have studied meniscus coating of tin perovskite films. One likely reason is due to the fast nucleation and crystal growth rate of tin perovskite [83], leading to uncontrollable crystallization, especially in meniscus coating.

Summary and outlook
The advancement of coating methods and morphology control strategies in perovskite film fabrication will undoubtedly bring us into a bright future of PV applications. It is clear that thoroughly understanding and control of perovskite solution spreading, drying, and crystallization during meniscus coating is essential to achieving device high-performance. Notably, in situ characterization of phase transition and morphology evolution has provided valuable insights into the dynamics of film formation. More comprehensive studies will further unveil the mechanisms of meniscus coating of perovskite films. Moreover, since meniscus coating is compatible with an R2R processing system, fundamental studies will guide the way to large scale, high-speed production. Nevertheless, numerous challenges remain. Comprehensive understanding of meniscus coating mixed cation/anion perovskite films must be attained. Lack of work in this arena see it lagging behind its spin-coating counterpart. Intensive studies of the film evolution in mixed cation/anion perovskite can provide direction to achieving high-quality films. Though the main issue covered in this review is meniscus coating the perovskite active layer, mass production of PSCs requires scalable manufacturing of each layer of the device stack. Polymer carrier transportation layers are the simplest challenge to overcome due to extensive study. The Achilles' heel is the scalable fabrication of the top electrode. The carbon electrode is a promising candidate and is more suitable to meniscus coating method than spin-coating. At this point, meniscus coating is the most promising technique to realize fully scalable fabrication of PSCs.