Microstructure Engineering of Metal-Halide Perovskite Films for Efficient Solar Cells Microstructure Engineering of Metal-Halide Perovskite Films for Efficient Solar Cells

Photovoltaic (PV) devices with metal-halide perovskite films, namely perovskite solar cells, have become a rapidly rising star due to low cost of raw materials, simple solu tion processability, and swiftly increased power conversion efficiency (PCE). The PCEs so far certified have gone beyond 22% for perovskite solar cells and 23.6% for tandem devices with single crystalline silicon solar cells, which offer a promising PV technology for practical applications. In principle, performance of perovskite solar cells are largely dominated by the optoelectronic properties and stability of metal-halide perovskite films, which are determined by the microstructure features of the films in turns. In this chap -ter, we will describe the recently developed strategies on microstructure engineering of metal-halide perovskite films for efficient perovskite solar cells. systematically understanding the role of microstructure engineering in the progress of perovskite solar cells. Dramatically pro moted crystallization control of organolead triiodide perovskite film by a homogeneous


Introduction
Photovoltaic (PV) devices with metal-halide perovskites that follow the formula of ABX 3 [A〓H 3 NH 3 (MA), NH〓CHNH 3 (FA), or Cs; B〓Pb, Sn; X = I, Br, and Cl] or their alloyed counterparts [1,2], namely perovskite solar cells, have swiftly emerged as a focal point of PV community owing to their low-cost, easy and large-scale fabrication, swiftly improved power conversion efficiency (PCE) to over 22% from 3.8% in just few years [3], and feasibility of tandem with well-developed PV technologies [4]. Those combined merits basically come from exceptional optoelectronic features of metal-halide perovskite films such as high absorption coefficient, tunable bandgap, long and balanced carrier diffusions, ambipolar transport of charge carriers, tolerance of defects [5,6], along with widely available raw materials and feasibility of film deposition via solution routes.
Critical to fabrication of efficient perovskite solar cells is deposition of high-quality metal-halide perovskite films with tailored microstructural features that mainly refer to surface coverage, grain size, texture, surface roughness, and so on [7,8]. All of them are previously revealed to play extremely important roles for ultima device performance and even stability. For example, as for surface coverage of metal-halide perovskite films, Snaith et al. [9] revealed that the lowest average efficiency is observed for the lowest coverage of MAPbI 3−x Cl x film, and vice versa. In terms of grain size of metal-halide perovskite films, Huang et al. [10] demonstrated that solvent annealing route can effectively increase the grain size and crystallinity of MAPbI 3 films, which resulted in much improved material electronic property and device performance. Similarly, Ohkita et al. [11] showed that the cells exhibited improved PV parameters with increase of grain size of MAPbI 3 films. In particular, reduced trap-assisted recombination resulted from large-sized grains gave rise to high open-circuit voltage (V oc ) for the cells. In addition, Park et al. [12] declared that as MAPbI 3 grain size decreases, the photocurrent density-voltage (J-V) hysteresis of perovskite solar cells tended to be more obvious. On the part of texture of metal-halide perovskite films, Jo et al. [13] successfully fabricated MAPbI 3 films with two different crystal orientations with respect to the substrate by using two different organic precursors (MAI and MACl). They found that the PV performance is directly related to crystal orientation of MAPbI 3 due to anisotropy of charge transfer in the crystal, suggesting that control of crystal orientation of metal-halide perovskites is important to realize high-performance cells. With regard to surface roughness, Kwok et al. [14] prepared a bilayer-structured MAPbI 3 film, which is compose of a mesostructured underlayer and a dense upper layer. They revealed that higher V oc of cell is directly related to smaller root mean squares (RMS) of MAPbI 3 upper layer. Overall, a metalhalide perovskite film with the desired microstructures of full surface coverage, large grain size, textured feature, and smooth surface, is highly desirable for efficient perovskite solar cells.
At this point, much research attention has been focused on microstructure engineering of metalhalide perovskite films. In the past year, there have been rapid progresses in this research field. Therefore, in this chapter we will focus our attention on the review of recently developed strategies on microstructure engineering of metal-halide perovskite films aiming to realize high-performance and stable perovskite solar cells. In particular, in the first section we begin with a general introduction to basic fundamentals of metal-halide perovskite materials that used as the absorbers for solar cells. In the second section, we come to the discussion of deposition technologies of metal halide perovskite films. And in the third section, we will discussion the recently developed strategies toward the microstructure engineering of metal-halide perovskite films by modifying the deposition technologies of metal halide perovskite films.

Basic fundamentals of metal-halide perovskites for solar cells 2.1. Crystal structures
The mineral perovskite named for Russian mineralogist Lev Perovski, possesses the chemical formula of CaTiO 3 . Compounds with the same crystal structure, i.e., ABX 3 are called perovskites. For a cubic unit cell, eight corners of the cube are occupied by A-cations and a B-cation is resided in its body center. The B-cation is encircled by six X-anions that located at the face centers of the cube unit cell. The six X-anions and one B cation form an octahedral [MX 6 ] 4− cluster. By changing the element types at A and/or B sites, we can obtain the perovskite variants with a variety of physical properties, such as semiconductive, superconductive, ferroelectric, antiferroelectric, etc. [16]. For the ones used for solar cells, the large cation at A site is commonly chosen to be organic cations such as methylammonium (MA, CH 3 NH 3 + ), formamidinium (FA, HC(NH 2 ) 2 + ), phenylammonium (PhA, C 2 H 5 NH 3 + ), or inorganic Cs + cation [17,18], as shown in Figure 1(a). A polar organic cation can yield a larger dielectric constant than an all inorganic system. However, organic cations are companied by poor stability of the compounds [19]. A large atom of Pb or Sn is generally settled down at B site, which then couples with halide X anions including Cl − , Br − , and I − to form metal-halide perovskite. By designing the component ratios, a series of metal-halide perovskite variants can be realized, which thus suggests an interesting and effective way to develop new materials [20]. In addition, replacing the halide at X sites with BF 4− , PF 6− , or SCN − can also engineer the materials' properties; especially in boosting their stability [17].
The structural formability and stability of metal-halide perovskites is guided by its geometric tolerance factor (t), t = (r A + r X )/[√2(r B + r X )], where r A , r B, and r X are the effective ionic radii of A, B, and X, respectively [21]. As indicated in Figure 1(b), for the ones with t close to 1, they prefers an ideal cubic structure, while they distorts into a low-symmetry structure, when t is smaller than 1. And, the ones with t value between 0.8 and 1.0 tend to adopt a cubic structure, and photoinactive non-perovskite structures are formed when the value of t is larger (>1) or smaller (<0.8) [22]. Figure 1(c) gives the estimated t of a series of metal-halide perovskites [23]. We can see that all the values are in the range of 0.8-1, which is a universal feature of perovskite structure. Replacing Pb with Sn appears to be beneficial to increase t; yet it induces the serious decrease of the compound stability. This mainly comes from the fact that t is not the only determinant factor, and Pb is inerter to oxidization than Sn [6]. In addition, as for organic cation, it seems that both charge distribution and size dominant the crystal structure. Previous works revealed that MA and FA benefit to stabilized perovskite structure, while the others with similar size, such as CH 3 CH 2 NH 3 + , (CH 3 ) 2 NH 2 + are easily induce the formation of a non-perovskite structure [22].
At different temperatures, a metal-halide perovskite crystal usually exhibits α, β and γ phases, as schematized in Figure 2 [24]. For example, the α to β phase transition for MAPbI 3 happens at 330 K, while the β to γ phase transition appears at 160 K. Noting that a non-perovskite δ phase was also found in some metal-halide perovskite variants such as HC(NH 2 ) 2 PbI 3 , FAPbI 3 , CsPbI 3 , and CsSnI 3 . The B-X bond of δ phase is broken. So it cannot be derived from the α phase by B-X-B bond angle distortion [25].

Electronic structures
Electronic band structures of metal-halide perovskites were initially investigated by Koutselas et al. [26] using band structure calculations. Then, Umebayashi et al. [27] studied the electronic band structures of cubic phase metal-halide perovskites based on ultraviolet photoelectron spectroscopy and first principles density functional theory (DFT) band calculation. It has been revealed that valence band maximum (VBM) states of MAPbX 3 and CsPbX 3 (X = Cl, Br, I) crystals contain Pb 6p-I 5p σ-antibonding orbital, while conduction band minimum (CBM) states compose of Pb 6p-I 5 s σ anti-bonding and Pb 6p-I 5p π antibonding orbitals [27]. Yan et al. [28] conducted the calculations of band structure, partial charge density of VBM and CBM states and (partial) density of states (DOS) of α phase MAPbI 3 . The results revealed that the direct bandgap locates at the R point. And, Pb has an occupied 6 s orbital below the top of valence bands. This lone pair of s electrons in MAPbI 3 maybe accounts for its unique optoelectronic properties. Further, partial charge density and DOS plots indicate that CBM largely compose of Pb p state, while VBM exhibits obvious Pb s and I p antibonding feature. This means that electronic structure of MAPbI 3 uniquely has the dual nature of ionic and covalent properties. In addition, the electronic states resulted from the organic cations are away from band edge. It means that organic cations have a negligible impact on basic electronic structures of MAPbI 3 [25].

Light absorption and bandgap
The feature of strong absorption over a wide range of spectrum is highly desired for absorber material of PV device, which is conductive to enlarge photocurrent, diminish the usage amount of raw material, and minimize the charge and energy loss during extraction to electrodes [6]. One highlighted property of metal-halide perovskite materials is their very high absorption coefficient along with sharp onset of absorption edge [2]. Moreover, most of metal-halide perovskite materials are direct bandgap semiconductors [28]. In terms of the most typical MAPbI 3 , its absorption coefficient exceeds 3.0 × 10 4 cm −1 in the visible range, which means that it only needs 380 nm thin film to absorb 90% of the visible light [2]. In fact, the thickness of light absorption layer for most of perovskite solar cells is around 300-600 nm. In addition, MAPbI 3 film shows a very sharp absorption onset, which means the low density of deep states of it. The Urbach energy of MAPbI 3 film was estimated to as low as 15 meV, which is comparable to monocrystalline GaAs [29]. The sharp absorption onset of MAPbI 3 maybe contribute to the small offset between its optical bandgap (~1.55 eV) and V oc of ultimate devices (~1.10 V) [6].
Another unique property of metal-halide perovskites is their bandgap tunability, which can be realized by substituting the atoms at A, B, or X sites. When MA was replaced with a slightly larger FA, the reduced bandgap from 1.55 to 1.48 eV was observed, corresponding to  an absorption edge extended from 800 to 840 nm, as firstly reported by Eperon et al. [30]. And, the photocurrent of the cell with FAPbI 3 film can reach up to 23 mA cm −2 , because of extended light absorption range. While a smaller Cs + at A site gives rise to an increased bandgap of 1.73 eV [31]. For X site, after replacing I − with Br − or Cl − , the bandgap increases to 2.30 eV for MAPbBr 3 or 3.12 eV for MAPbCl 3 , respectively. Figure 3(a-c) shows the continuous bandgap variation of FAPbI y Br 3−y (0 < y < 1) films. Its absorption onset changes from 840 nm for FAPbI 3 to 550 nm for FAPbBr 3 , corresponding the bandgap increases from 1.48 to 2.13 eV. For B site, a smaller cation size usually results in less bandgap. For example, the bandgap of MAPbI 3 decreases to 1.2 eV by replacing the Pb + with Sn + [24]. Similarly, the bandgap decreases from 1.67 eV for CsPbI 3 to 1.30 eV for CsSnI 3 and 1.08 eV for CsGeI 3 [24].

Long-range ambipolar charge transport
For most of semiconductors, the transport of electrons and holes is unbalanced, due to different effective masses (m e * and m h * ). Surprisingly, numerous independent experimental studies have revealed that well-balanced electron and hole transport is established in metalhalide perovskites, namely ambipolar transport property. First principle calculation also indicates that this fact that electron effective mass (m e * = 0.23 m 0 ) of this type materials is extremely similar to hole effective mass (m h * = 0.29 m 0 ) [32]. Balanced ambipolar transport has an effect on bulk polarization during charge transport and collection; in turns affects PV parameters of perovskite solar cells. Moreover, metal-halide perovskites can convey both n-type and p-type properties when they are used in thin-film devices with different interfacial layers [22]. In addition, long carrier diffusion length of ~100 nm and ~1 μm were revealed for solution-processed MAPbI 3 and MAPbI 3−x Cl x films, which are much larger than that of organic semiconductors. Almost certainly, long diffusion length can bring reduced charge recombination and improved charge collection, thus excellent cell performance [18].

Exciton binding energy
The hybrid nature of metal-halide perovskite materials arise wonders whether incident photons generate free charges or bound excitons. So, the exciton binding energy of metal-halide perovskite materials has been intensively concerned [18]. In the 1990s, theoretical studies have revealed the exciton binding energy to be around 37 meV for MAPbI 3 [33]. This value is much lower than that of organic materials with the values ranged from 0.2 to 1 eV. The low binding energy together with long diffusion length maybe account for the high PECs of perovskite solar cells. The low binding energy of metal-halide perovskites is mainly related to their large dielectric constants, which give rise to strong dielectric screening effect [18]. Recently, Hu et al. [34] found that the dielectric constant MAPbI 3 is as high as ~32 under dark condition. And, the value can be greatly enhanced (~1000) under photoexcitation as investigated by Juarez-Perez et al. [35]. The high dielectric constant gives rise to strong dielectric screening effect, thus low exciton binding energy.

Film formation technologies
Up to now, various processing techniques have been developed to prepare metal-halide perovskite films, as summarized in Figure 4. They mainly include one-step spin-coating method, sequential solution deposition method, two-step spin-coating method, vacuum coevaporation deposition method, sequential vacuum deposition method, and vapor-assisted solution deposition method [36].
The one-step spin-coating method involves spin-coating of a precursor solution containing PbX 2 with a certain amount of MAX firstly (X = I, Br, and Cl). Then, the metal-halide perovskites formed and grew during solvent evaporation. A post-annealing recipe with a temperature of ~100°C was usually required to remove residual solvents and complete crystallization. For this method, the film morphology and quality strongly depend on the processing conditions such as annealing temperature, solution concentration, precursor composition, solvent choice, etc. [36]. Although it is extremely simple, the one-step spin-coating method faces the difficulty to the deposition of pinhole-free metal-halide perovskite films. Solvent engineering is proved to be one of effective routes to overcome this obstacle. Spiccia et al. [37] proposed a fast deposition crystallization method to induce the crystallization of MAPbI 3 during spin-coating process. This method includes the spin-coating of MAPbI 3 precursor with N,N-dimethylformamide (DMF) as the solvent, followed by dropping anti-solvent such as toluene and chlorobenzene to complete the crystallization of MAPbI 3 . The anti-solvent decreased the MAPbI 3 solubility in DMF solvent, and thereby promoting fast nucleation and crystallization [6]. Later, Jeon et al. [38] designed a mixed γ-butyrolactone and dimethyl sulfoxide (DMSO) as the processing solvent followed by toluene drop-casting. The difference is that a dark brown MAPbI 3 film formed immediately after dropping of toluene solvent, while a transparent MAI-PbI 2 -DMSO intermediate phase formed firstly after dropping of toluene using the mixed solvent. Both X-ray diffraction (XRD) and Fourier-transform infrared spectroscopy (FTIR) measurements verified the formation of intermediate phase. After being thermal annealed at 100°C for 10 min, intermediate phase film can transform to an extremely uniform and compact MAPbI 3 layer [6]. Similarly, Park et al. [39] further developed the intermediate phase route based on Lewis acid adduct of PbI 2 , as shown in Figure 5. This method eventually resulted in an average PCE of 18.3% from 41 cells and the best value of 19.7%.
The sequential deposition procedure was firstly introduced by Grätzel et al. [40]. And in surprise, it yielded a high PCE of 15%. This method includes the spin-coating of a PbX 2 solution on substrate, followed by drying, and then dipping the substrate into an isopropanol (IPA) solution of MAX to accomplish the reaction in few minutes. Typically, the solutions are 1 M PbI 2 in DMF and MAI in IPA with relative low concentration from 0.004 to 0.006 M. After the dipping step, the residual MAI was rinsed out by IPA. With this method, the films' homogeneity can be dramatically improved and their morphology becomes more controllable than the one-step spin-coating method.
The two-step spin-coating method that is also known as interdiffusion method was proposed by Huang et al. [41]. It can be seen as the one derived from sequential deposition procedure. But, the difference is that metal-halide perovskite is formed by spin-coating an MAI layer on PbX 2 precursor film followed by thermal annealing at 100°C for a relative long time up to 2 h. This method is materials saving, and can synthesize more uniform films than sequential deposition.
Snaith et al. [42] firstly reported the vacuum co-evaporation deposited planar MAPbI 3−x Cl x perovskite solar cells by co-evaporating PbCl 2 and MAI in a vacuum thermal evaporation system. The PCEs with a narrow distribution and optimal one of 15.4% were observed in contrast with that of one-step solution process. Due to its low control accuracy in precursor ratio, the sequential vacuum evaporation technology was further developed via depositing PbCl 2 and MAI layer by layer [43]. By heating the substrate during precursor evaporation, a pure-phase, homogeneous, and pinhole-free CH 3 NH 3 PbI 3 − x Cl x film can be synthesized. However, the requirement of high vacuum degree of them inevitably increases the device cost, and thus partially restricts their practical applications.
Vapor-assisted solution deposition is firstly used by Yang et al. [44]. For it, PbI 2 film that was pre-deposited via spin-coating route reacted with MAI vapor under atmospheric pressure. The resultant MAPbI 3 film was revealed to be extremely smooth, and composed of microscale polycrystallites, delivering to a PCE of 12%.

Microstructure engineering of metal-halide perovskite films
The microstructural features of metal-halide perovskite films such as surface coverage, grain size, texture, surface roughness, and so on are previously revealed to play extremely vital  roles for determining ultima device performances and even stability. In the past year, there have been rapid progresses in the research field of microstructure engineering of metal-halide perovskite films. In this section, we will focus our attention on the recently developed strategies on microstructure engineering of metal-halide perovskite films aiming to realize highperformance and stable perovskite solar cells.

Surface coverage engineering of metal-halide perovskite films
The reasons for ensuring surface coverage of metal-halide perovskite film in perovskite solar cells come from the following two reasons. On the one hand, if there are some regions without metal-halide perovskites padding, light will travel directly without absorption, which leads to decreased photocurrent in turns. On the other hand, any existed pinholes inevitably result in direct contact of electron transport layer with hole transport layer, thus resulting in the formation of shutting paths. They will form additional parallel resistors, causing declines in performance parameters of the cell.
The traditional one-step spin-coating method faces the difficulty to yield a uniform and fullcoverage metal-halide perovskite film in large areas. Aiming to overcome this issue, some effective modifications have been reported. For example, as given in Figure 6(a), Yu et al. [45] introduced a recrystallization process via DMF vapor fumigation to induce the self-repair of one-step deposited MAPbI 3 films with poor coverage and low crystallinity. By adjusting the cycle of recrystallization process, they found that MAPbI 3 films with dendritic structures spontaneously transformed to the uniform ones with full coverage and high crystallinity (Figure 6(b-e)). Solar cells with these modified MAPbI 3 films yielded reproducible average PCE of 10.25 ± 0.90% and the optimal one of 11.15%, which is both much higher than that of non-modified MAPbI 3 films (Figure 6(f-d)). In addition, the J-V hysteresis in the measurement of cell performance can also be effectively alleviated. The authors attributed this desired feature to improve the quality of MAPbI 3 films in the optimized devices.
Cui et al. [46] discovered that methylamine (CH 3 NH 2 ) gas can trigger defect-healing of MAPbI 3 films via room-temperature ultrafast, reversible chemical reaction of MAPbI 3 with CH 3 NH 2 gas. They revealed that healing of MAPbI 3 films can be ascribed to the formation and reconstruction of an intermediate MAPbI 3 ·xCH 3 NH 2 liquid phase during perovskite-gas interaction. MAPbI 3 film processed by one-step spin-coating method using DMF as solvent is composed of dendrite-like MAPbI 3 crystals. And voids with size up to several micrometers between them can be clearly found. After CH 3 NH 2 induced defect-healing treatment, dendrite-like crystals and voids almost disappeared and a dense, smooth MAPbI 3 film has formed. And, AFM measurement further revealed that the healed film has a very dense and smooth surface, with a RMS roughness of ~6 nm. Benefiting from improved surface coverage of MAPbI 3 films by methylamine-induced defect-healing, obvious increase of PCE from 5.7 to 15.1% were realized for the cells. Afterwards, this interesting chemical reaction of CH 3 NH 2 gas with metal-halide perovskite was investigated in detailed and extended for further uses such as reduction of intrinsic defect concentration of MAPbI 3 films [47], realization of solventand vacuum-free deposition of MAPbI 3 films [48], and so on.

Grain size engineering of metal-halide perovskite films
Increasing theoretical and experimental evidences indicate that, similar to well-developed thin-film PV devices such as CdTe and Cu(InGa)Se 2 , primary energy loss in perovskite solar cells is also ascribed to non-radiative recombination of carriers at undesirable trap states. In general, for polycrystalline perovskite films trap states mainly come from crystal imperfections especially such as grain boundaries and intragranular defects [49,50]. While number of grain boundaries is inversely proportional to average grain size for polycrystalline film, so numerous works have been focused on increasing grain size of metal-halide perovskite films by modifying deposition technologies of metal halide perovskite films.
For example, as shown in Figure 7, Yu et al. [49] demonstrated that a homogeneous cap-mediated crystallization with face-to-face configuration can control the crystallization kinetics of MAPbI 3 films in one-step spin-coating method. They found that homogeneous caps, especially the ones with low surface roughness, can effectively retard the nucleation rate, promote growth, and prevent composition loss of MAPbI 3 grains. Thus, pinhole-free MAPbI 3 films can be formed, which have many desirable features, such as greatly enlarged grains, significantly improved crystallinity, preferred (110) orientation, vertically aligned grain boundaries, and proper stoichiometry. As a result, planar-resultant heterojunction solar cells yielded a much enhanced average PCE of 17.87%. It should be noted that large fill factors (FFs) were observed in these efficient cells. In subsequent work, they revealed that PbI 2 heterogeneous cap can also realize MAPbI 3 films with large-sized grains [51]. Improved PCE was thus realized because of more efficient transport of charge carriers and decreased non-radiative recombination in corresponding devices. Overall, those works suggest a promising strategy to engineer grain size of metal-halide perovskite films.
In addition to crystallization process control, post-treatment strategies were also developed to engineer grain size of metal-halide perovskite films. For example, obvious grain coarsening via Ostwald ripening in one-step deposited MAPbI 3 film can be realized by post-synthesis high-temperature heating treatment assisted with additionally deposited CH 3 NH 3 I layer [50]. The grain coarsening via Ostwald ripening was revealed to be related to the heating treatment parameters (temperature and time). By optimizing them, the film with average grain size of ~2 μm, much increased crystallinity, and proper stoichiometry can be achieved. Due to those characteristics, defect states along with recombination centers were greatly reduced, and carrier transport and injection properties were much improved. So, efficiency of corresponding planar heterojunction solar cells can be boosted from 14.54 to 16.88%. Then, the same post-treatment recipe was used for thick MAPbI 3 films, and a same grain coarsening was observed in them. So post-treatment recipe gives the fact that thickening the absorb layer of cells to realize more sufficient absorption avoids serious aggravation of charge recombination. By further optimizing the thickness of coarsened MAPbI 3 films, highly efficient cells with relatively excellent reproducibility and the optimal efficiency of 19.24% were realized by Yu. et al. [50]. Afterward, a similar MABr treatment converts MAPbI 3 thin films to high-quality MAPbI 3−x Br x thin films following an Ostwald ripening process as reported by Zhao et al. [52]. But, they found that similar process is ineffective when replacing MABr with MAI. This phenomenon mainly comes from the fact that low-concentration MAI solution was used and low post-treatment temperature was adopted in their experiments. So, further investigations are needed to clarify those factors. More recently, Jen et al. [53] reported a simple pseudohalideinduced film retreatment technology as passivation for preformed MAPbI 3 film. They found that the retreatment process yields a controllable decomposition-to-recrystallization evolution of MAPbI 3 film. Corresponding, it remarkably enlarges grain size of the film in all directions, as well as improving crystallinity and hindering trap density.

Texture engineering of metal-halide perovskite films
As to polycrystalline films, orientation of crystal axis in each grain is another important microstructural feature that dominates their physical properties. Films with aligned crystal axes are so-called textured ones. They possess a single-crystal-like nature along crystal axis, so an enhancement in physical properties is expected for them. In general, ordinarily prepared polycrystalline films are composed of grains with random orientation. Methodology that is explored to develop texture to improve functional properties of polycrystalline films is known as texture engineering. Specifically, one-step deposited metal-halide perovskite films are similarly characterized with randomly oriented grains. Hence, texture engineering is of particular importance to modify their electrical and optical properties, and hence further improve the performance of ultimate cells.
Yan et al. [54] reported that reaction of HPbI 3 with low partial pressure MA gas can form a textured MAPbI 3 film with high crystallinity. The film exhibits much higher both thermal and moisture stability than the one prepared from MAI + PbI 2 . Further investigation revealed that large Pb-N binding energy of ~80.04 kJ mol −1 results in a liquefied state after MA adhesion on MAPbI 3 . And, a highly textured MAPbI 3 film is formed when excess MA expeditious are released. Cao et al. [55] found that MACl-containing precursor can yield MAPbI 3 film with strong (110) preferred orientation. The MAPbI 3 films were used for typical planar solar cells and delivered an impressive average efficiency of 16.63 ± 0.49% and champion efficiency of 17.22%. Yu et al. [56] demonstrated that face-down annealing of one-step deposited precursor films can enable the formation of (110) textured MAPbI 3 films consisting of high-crystallinity, well-ordered, micrometer-sized grains that span vertically the entire film thickness, as shown in Figure 8. Such microstructural features induced dramatically decreased nonradiative recombination sites as well as greatly improved transport property of charge carries in the films compared with that of the non-textured ones obtained by conventional annealing route. As a consequence, planar heterojunction perovskite solar cells with these textured MAPbI 3 films exhibit much improved PCE along with small hysteresis and excellent stability.

Surface roughness engineering of metal-halide perovskite films
Metal-halide perovskite film was usually sandwiched between electron-transporting layer and hole-transporting layer in perovskite solar cell. And, one of them has to be deposited sequentially on metal-halide perovskite layer. So, surface roughness of metal-halide perovskite film has a significant impact on the cell's interface morphology. A roughened interface that resulted from rough metal-halide perovskite layer would strengthen internal light scattering [57]. And, a large interface area also benefits charge transport [14]. However, high surface roughness of metal-halide perovskite layer will increase short-circuiting possibility of device existing between silver electrode and metal-halide perovskite layer, in the case that electron-transporting layer or hole-transporting layer cannot fully cover the metalhalide perovskite film. In other words, a metal-halide perovskite layer with high roughness requires a thick electron-transporting layer or hole-transporting layer to eliminate shortcircuiting. On the contrary, a thin one is preferred to ensure a reasonable FF. So, there is compromise in surface roughness of metal-halide perovskite film as far as cell performance is concerned, and some exploratory works have been undertaken. For example, one the one hand, Meng et al. [58] introduced a hot-pressing method that can transform MAPbI 3 film with rough surface to be a smooth one, and pinholes in original film can be cured effectively. This modified MAPbI 3 morphology is conductive to improve charge carrier transport and eliminate charge carrier recombination in perovskite solar cells. Moreover, much improved performances with high PCEs of 10.84 and 16.07% are thus realized in hole-transporting-layer-free and spiro-OMeTAD-based cells, respectively. Yu et al. [59] found that spray-assisted process instead of commonly used dipping process in a two-step spin-coating method can effectively reduce roughness of MAPbI 3 films. Experimental results demonstrated that the cells' average V oc can be enhanced from 0.823±0.105 V to 0.940±0.008 V by spray-assisted process. It benefits from low leakage possibility between hole-transport layer and mesoporous TiO 2 layer when a smooth and pinhole-free MAPbI 3 has successfully formed. Finally, average PCEs of mesoporous cells could be promoted by 25% approximately. One the other hand, Chen et al. [60] reported a novel MAPbI 3 film with a dense under-layer and a porous upper-layer that was formed by using a thin mesoporous TiO 2 seeding layer and a gas-assisted crystallization method. This novel multitiered nanostructure allows for greatly improved light harvesting for wavelengths exceeding 500 nm, as well as a more effective interfacial charge separation for perovskite solar cells. The combination of these factors culminated in average PCEs over 15% and average short-circuit current density exceeded 22 mA cm −2 .

Conclusions
In summary, metal-halide perovskite films have many excellent optoelectronic properties such as high absorption coefficient, tunable bandgap, long and balanced carrier diffusions, ambipolar transport of charge carriers, tolerance of defects, along with capacity for film deposition via either solution or vacuum-based methods including one-step spin-coating method, sequential deposition method, two-step spin-coating method vacuum coevaporation deposition method, sequential vacuum deposition method, and vapor-assisted solution deposition method. Those desired features make them promising for high-performance and low-cost perovskite solar cells. The microstructural features that mainly refer to surface coverage, grain size, texture, surface roughness, and so on are vital in determining the performance of perovskite solar cells. Specifically, the ones with full surface coverage, large grain size, textured feature, and smooth surface, are highly desirable for efficient devices. Some important progresses in microstructure engineering of metal-halide perovskite films are described in this chapter, which will promote systematically understanding the role of microstructure engineering in the progress of perovskite solar cells.