Spectroscopic ellipsometry (SE) was used to investigate the role of isopropyl alcohol (IPA) solvent in the synthesis of organic-lead-halide perovskite CH(NH2)2PbIxBr3−x [FAPbIxBr3−x] thin films including the effect of I/Br composition ratio by the two-step reaction of an amorphous (a-)PbIxBr2−x layer and FAIxBr1−x solution diluted in IPA. An optical dispersion model was developed to extract the complex refractive index N (=n + ik), optical transition, and film thickness of FAPbIxBr3−x perovskites by SE analysis at different I/Br composition ratio as a function of immersion time in a solution of FAIxBr1−x diluted in IPA. SE combined with X-ray diffraction (XRD), Fourier-transform infrared spectroscopy (FTIR), and scanning electron microscopy (SEM) revealed that Br incorporation into films promoted the densification of FAPbIxBr3−x perovskite network along with increased film thickness and volume fraction of void. IPA promoted film crystallization of a-PbIxBr2−x accompanied by the formation of surface roughness, grain boundaries, and voids, followed by enhanced diffusion of FAIxBr1−x into the grain boundaries/voids in the mesoporous crystallized PbIxBr2−x network. These processes contribute synergistically to the growth of the perovskite structure.
- organic-lead-halide perovskite
- two-step reaction
- role of isopropyl alcohol (IPA) solvent
- growth mechanism
Ragavan In recent years, organic-inorganic halide perovskite solar cells have significantly attracted much interest as high-efficiency solar cell material due to their high absorption coefficient, ambipolar carrier transport, and long-carrier diffusion lengths, which have resulted in power conversion efficiencies (PCEs) as high as 20.1% [1–3]. Methylammonium [(CH3NH3)+; MA] lead triiodide (MAPbI3), which is used in the most efficient perovskite solar cells, has a bandgap (Eg) of around 1.57 eV [4, 5], a very sharp absorption edge , and an impressively low difference between the open-circuit voltage (Voc) and its bandgap potential (Eg/q) [6–8]. However, the instability of MAPbI3 with respect to humidity, temperature, and light soaking is a serious problem hindering long-term reliability of the photovoltaic performance. Therefore, the formamidinium [(NH2)2CH+; FA] lead triiodine (FAPbI3) perovskite, where the ionic radius of the A-site FA+ ion (253 pm) is larger than that of MA+ (217 pm), has recently been extensively studied due to better-phase stabilization and higher photovoltaic performance over other systems such as MAPbI3, and better stability for air storage with an extension of the optical absorption edge to the infrared region up to around 840 nm . More recently, interest has extended to the FAPbIxBr3−x and Cs0.2FA0.8PbIxBr3−x perovskite, which has even higher stability against humidity as well as temperature. A bandgap is also tunable from 1.43 to 2.35 eV by substitutional doping of bromide ions to iodine sites .
In addition, various methods for the fabrication of perovskite thin films have been extensively studied, such as a one-step solution process , sequential two-step reaction [11, 12], and vacuum deposition . Although an efficient PCE of around 20% has been achieved using uniform and dense perovskite layers prepared by a one-step solution process [14, 15], this involves complex procedures such as precise control of the composition ratio in the precursor solution and the rate of evaporation to adjust the composition ratio of the perovskite film. An alternative approach is a two-step deposition method using porous PbX2 (X = I, Cl, Br) films and organic halide solutions such as MAX and FAX diluted in 2-isopropanol alcohol [IPA; (CH3)2COH] solvent on an electron transport layer such as mesoporous TiO2 or ZnO, which results in better coverage and uniformity than the one-step process. However, the film morphology of the porous PbX2 in the two-step deposition method has a strong effect on the chemical reactivity with the alkyl amine halide, which in turn determines the film homogeneity and crystallinity of the resultant perovskite thin films. To control the morphology of PbX2 films, solvent vapor treatment , and additives such as hydriodic acid , dimethyl sulfoxide (DMSO), and N-methyl-2-pyrrolidone  have been tested. For example, Miyadera et al. reported that the rate of perovskite formation from polycrystalline PbI2 on mesoporous TiO2 was approximately one order of magnitude faster than that of well-oriented dense PbI2 on glass . In addition, the type of solvent is sensitive to the morphology of PbI2. N,N-Dimethylformamide (DMF; CH3NCHO) is generally used to promote the crystallization of PbI2. The use of DMSO or cumene hydroperoxide (CHP; C6H5C(CH3)2OOH), which has lower vapor pressures than DMF, has been reported to suppress the rapid crystallization and to form an amorphous structure that reacts preferentially with MAI diluted in IPA solvent, resulting in a slower rate of crystallization of MAPbI3 and a more homogeneous film structure [20, 21]. However, the role of the polar IPA solvent in the synthesis of FAPbIxBr3−x perovskites is still not clear because the reaction rate is very fast. Thus, to understand the effect of Br incorporation on the fine structure and the growth mechanism of perovskites by a two-step method from PbIxBr2−x film and FAIxBr1−x solution diluted in IPA solvent, a slower reaction rate is preferable from amorphous (a-)PbIxBr2−x rather than from mesoporous crystalline PbX2 by immersion in FAX/IPA solution at lower temperatures.
In this chapter, we present the role of the IPA solvent in the synthesis of FAPbIxBr3−x perovskites including the effect of Br incorporation in the films by a two-step solution process using spectroscopic ellipsometry (SE) as well as X-ray diffraction (XRD), Fourier-transform infrared spectroscopy (FTIR), and scanning electron microscopy (SEM) observation.
2. Experimental section
A-PbIxBr2−x thin films with a film thickness of around 150 nm were prepared on glass by spin coating a mixture of FAIxBr1−x and PbIxBr2−x powder (PbI2:PbBr2 = 3:1 and 3:2 molar ratio) in a DMF/CHP (95:5, v/v) cosolvent at 3000 rpm for 30 s. A-PbIxBr2−x films prepared using the DMF/CHP cosolvent promoted the growth of perovskite structure after immersion in FAIxBr1−x/IPA solution, whereas no reaction to form the perovskite occurred in a-PbIxBr2−x films prepared from 100 vol% DFM solvent. Thus, a-PbIxBr2−x fabricated with DMF/CHP cosolvent was used as a precursor to promote the formation of perovskite structure. A-PbIxBr2−x films were immersed in a solution of FAIxBr1−x (FAI:FABr = 3:1 molar ratio) in IPA solvent for various times (tim) at 25°C. Finally, the FAPbIxBr3−x films were rinsed with IPA to remove the residual solution and contaminants from the film surface.
The refractive index n and extinction coefficient k of the corresponding FAPbIxBr3−x films were determined using a phase-modulated spectroscopic ellipsometer (UVISEL, Horiba Jobin Yvon). The ellipsometric angles, Ψ and Δ, which determine the complex reflection coefficient ratio, , were also measured for 71 points in the range of 1.5–5.0 eV (0.05 eV step) with an integration time of 200 ms at each photon energy. The angle of incidence, θi, of the probe light was 70°. The measured values for the n and k spectra, denoted by <n> and <k>, include information regarding the surface roughness, voids, and bulk inhomogeneity of the perovskite film and the PbIxBr2−x precursor. The bulk component of optical constants for the perovskite films, n and k, and the volume fractions of perovskite and voids (fperov, fvoid) were determined using a modified amorphous model based on the Forouhi-Bloomer (FB) model described in Eq. (1), combined with the Bruggeman effective medium approximation (EMA) [22, 23]:
The quantities Aj, Γ0, ωj, and ωg represent the resonance amplitude, the oscillator-broadening parameter, the resonance energy, and the Tauc optical gap, respectively. Spectral fitting was performed using the optical constant for bulk FAPbIxBr3−x perovskite described in Eq. (1), and fperov, fvoid, and each film thickness di (i = 1, 2, 3), as variables. A probable structure was determined using the optical model by minimization of the standard deviation χ2 between the measured and calculated ellipsometric error parameters using a linear regression method as follows :
where the superscripts c and m represent the calculated and measured values, and N and M are the numbers of the measured and calculated wavelengths, respectively.
The fine structures of the a-PbIxBr2−x and perovskite FAPbIxBr3−x thin films were also evaluated using X-ray diffraction (Bruker D8 ADVANCE ECO) with the diffractometer in Bragg-Brentano mode using Cu Kα radiation with a Lynxeye XE detector, Fourier-transform infrared spectroscopy (Shimadzu spectrometer, IRTracer-100), field emission scanning electron microscopy (FESEM; S4800, Hitachi High Technologies, Japan) with energy-dispersive X-ray spectroscopy (EDX) (Bruker XFlash 5030/Quantax 400), and atomic force microscopy (AFM; NanonavieII/SPI-3800, Hitachi High-Tech Science, Japan) to support the results of the SE analysis.
3. Results and discussion
3.1. Characterization of FAPbIxBr3−x perovskite films
Figure 1a shows XRD patterns of corresponding FAPbIxBr3−x films on glass synthesized by the two-step method using a-PbIxBr2−x layer and FAIxBr1−x solution in IPA at various tim. No XRD diffraction peaks that would indicate a crystalline phase were observed for spin-coated PbIxBr2−x, which suggests that the film structure was mainly amorphous. However, the film immersed for 10 s showed a diffraction peak at 2θ = 12.6°, which is attributed to the (1 0 0) crystal orientation in PbIxBr2−x, and peaks were observed at 2θ =14.6° and 28°, which correspond to the (1 1 0) and (2 2 0) diffraction peaks of the FAPbIxBr3−x perovskites, respectively. A trace of the hexagonal structure δ phase was also indicated by the peak at 2θ = 11.7°, which is undesirable with regard to the photovoltaic performance. The diffraction peaks attributed to the FAPbIxBr3−x perovskites increased with suppression of the δ phase peaks and the diffraction peak at 2θ =12.6° attributed to the (1 0 0) crystal orientation in PbIxBr2−x when tim was increased up to 600 s. The AFM image of 10 and 600 s immersed film is also shown in Figure 1b. RMS is increased from 28 to 63 nm. These results suggest that the I/Br composition ratio determines the fine structure, and the formation of a perovskite structure is promoted together with an increase of the surface roughness, grain boundaries, and voids.
3.2. SE study of perovskite FAPbIxBr3−x films
Figure 2 shows <n> and <k> spectra for the FAPbIxBr3−x perovskite films prepared at various tim, including those for the a-PbIxBr2−x precursor film. Once the a-PbIxBr2−x layers were immersed in the FAIxBr1−x solution in IPA, the <n> value decreased gradually, which suggests an increase in porosity, voids, and/or surface roughness in the a-PbIxBr2−x structure. By contrast, the <k> value that corresponds to optical absorption () in the visible to near-infrared region was markedly increased for the film immersed for 10 s, followed by a decrease with tim up to 600 s. Thus, the time evolutions of <n> and <k> spectra with tim include information of the growth chemistry of the FAPbIxBr3−x perovskite structure by the two-step process of a-PbIxBr2−x layer formation and immersion in the FAIxBr1−x solution in IPA.
Figure 3a shows the <n> and <k> spectra for the spin-coated a-PbIxBr2−x layer included in Figure 2 together with the best-fit calculated spectra. A single-layer model consisting of a modified FB model was used to determine the optical constant of the bulk component of the a-PbIxBr2−x layer and the film thickness. The best-fit <n> and <k> spectra are also included as dotted lines. Good-fitting results were obtained over the entire energy region from 1.5 to 5 eV with χ2 of 0.71. The n and k spectra of the a-PbIxBr2−x bulk component along with the optical model used for the spectra fitting are shown in Figure 3b. The film thickness was 153 nm, which was almost consistent with that determined by FESEM observation and a thickness profiler. The fine structures were observed at 3.1, 3.7, and 4.2 eV in a-PbIxBr2−x, which are analogous to those at 2.95, 3.50, and 3.9 eV attributed to the respective Γ4+ → Γ4−(pxy), Γ4+ → Γ4−(pz), and A4+ → A4− (conduction band) optical band transitions in PbI2 . These high-energy shifts in the fine structures are due to the addition of Br.
To understand the formation mechanism of perovskite structure using a two-step method, the measured <n> and <k> spectra were analyzed using two- or three-layer models consisting of a perovskite FAPbIxBr3−x top layer and a PbIxBr2−x underneath layer using constant (n, k) spectra for FAPbIxBr3−x and PbIxBr2−x, and fperov, fPb(I,Br)2, fvoid, and di as variables. However, the (n, k) spectra calculated using several optical models deviated far from the measured spectra, which suggests that modelization of the optical constant for the transitional stage from a-PbIxBr2−x to the FAPbIxBr3−x perovskite structure should be considered at each tim. Among single-layer to four-layer optical models, the three-layer model showed the best fit with four Lorentzian oscillators for a PbIxBr2−x and FAPbIxBr3−x composite phase including fperov, fvoid, and di (i = 1, 2, 3) as variables.
Figure 4a shows the best-fit n and k spectra for the FAPbIxBr3−x perovskite bulk component prepared at various tim. The n value increased in the first 10 s and then decreased with increasing tim, whereas the k values increased with the appearance of several fine structures attributed to the optical band transition of FAPbIxBr3−x. The absorption edge shifted to a lower energy from 3.1 to 1.63 eV for 60-s immersion and then became almost independent of tim. In addition, fine structures at 2.55 and 3.8 eV were shifted to the infrared region with an increase in their amplitude when tim was increased up to 600 s. In particular, the layer immersed for 600 s exhibited four fine structures in the k spectra at 1.63, 2.52, 3.25, and 4.79 eV, which were attributed to the optical transition of the perovskite structure.
The best-fit optical models for the various tim are summarized in Figure 4b, including the corresponding χ2 values. For the a-PbIxBr2−x films immersed for 10 and 60 s, no significant differences in fperov and fvoid were observed between the two- and three-layer models. The fperov and fvoid values were almost identical for both the top and intermediate layers, despite the use of the three-layer model. A prominent feature was that fvoid was increased significantly for the entire film thickness, despite being immersed for only 10 s, which suggests that the rate of crystallization of a-PbIxBr2−x was much quicker than the rate of formation of the perovskite structure. The diffusion of FAIxBr1−x into the voids and/or grain boundaries resulted in the simultaneous formation of large amounts of free volume and an increase in the film thickness, resulting in a decrease in the refractive index (Figure 4a). The best-fit d (=d1 + d2 + d3), d1, and fvoid for the top layer are summarized as a function of tim in Figure 5c, together with those determined using a thickness profiler. The value of d increased from 150 nm for a-PbIxBr2−xto 350 nm for the FAPbIxBr3−x perovskites. The volume fraction of voids and the thickness of fvoid and d1 corresponding to the top layer increased significantly during the initial 10–60 s of immersion, after which they increased only gradually with tim. These results imply that the formation of the perovskite structure proceeds with the formation of free space within the PbIxBr2−x network due to the insertion of the organic FA cation.
Figure 5a shows the n and k spectra for the FAPbIxBr3−x perovskite bulk component after immersion for 600 s, including each of the optical transition components determined by the deconvolution method. The electronic structure of MAPbI3 determined from DFT calculations was reported by Shirayama et al.  (Figure 5b). The comparison between optical transition energies for ω0, ω1, ω2, and ω3 of MAPbI3 and FAPbIxBr3−x is summarized in Figure 5c [26, 27]. Similar to the band structure of MAPbI3, four distinct optical transitions in FAPb(IxBr1−x)3 were observed at 1.63, 2.52, 3.25, and 4.79 eV, which may be attributed to direct semiconductor-type transitions at the R, M, X, and Г points in the pseudo-cubic Brillouin zone, respectively. According to Shirayama et al., a sharp optical transition at 3.25 eV in PbIxBr1−x is probably excitonic, and the transition energy of this peak is close to that of the peak for MAPbI3. The absorption peak at 3.1 eV in a-PbIxBr2−x is relatively sharp, while the corresponding peak was broad for PbIxBr2−x after immersion for 10 s. Two distinct optical transitions at 1.77 and 2.5 eV were observed for the film immersed for 60 s. Here, the absorption peak at 2.5 eV is probably attributed to a layered structure of PbIxBr2−x analogous to that of PbI2 reported elsewhere . The onset of the optical transition (ω0) was shifted to 1.75 eV and was almost independent of the immersion time over 300 s.
Figure 6 shows the changes of the optical transition energies, ω0, ω1, ω2, and ω3 with tim. The optical transition energies ω0, ω1, and ω2 shift to the infrared region and their amplitude increases when tim was increased up to 600 s. Among the four optical transitions, the lower-energy shift of the ω2 component from 3.85 to 3.4 eV was prominent and attributed to the 6s-6p transition of Pb2+. Hirasawa et al. reported the exciton features in zero- (0D), two- (2D), and three-dimensional (3D) networks of [PbI6]4− octahedra in MA4PbI6,2H2O, (C10H21NH3)2PbI2, and MAPbI3 from reflection spectra measured at 4 K . The energy of the lowest exciton decreases significantly as the dimension is increased and was 3.4, 2.55, and 1.633 eV for the 0D, 2D, and 3D network compounds, respectively. These results make an analogy that the process for the formation of the FAPbIxBr3−x perovskite from a-PbIxBr2−x is attributed to a decrease in the dimension of the crystal structure based on [Pb(IBr)6]4−.
3.3. Effect of x on the fine structure of FAPbIxBr3−x perovskite films
Figure 7 shows measured Is (=sin2ψsinΔ) and Ic (=sin2ψcosΔ) spectra for FAPbIxBr3−x perovskite films with two different PbI2:PbBr2 molar ratios of 3:1 and 3:2, including the best-fitted calculated spectra using the optical model shown on the right. The n of an I:Br = 3:2 film was larger than that of 3:1 film with increased film thickness and fvoid. Thus, increased Br incorporation into FAPbI3 promoted the densification of perovskite network, resulting in the increased film thickness and fvoid, as far as a two-step method with immersion to FAIBr/IPA solution was used. In fact, XRD pattern and PL revealed that lattice parameter decreased with higher-edge emission peak energy when Br composition was increased. SE also showed an increase of the refractive index for FAPbIxBr3−x perovskite phase together with an increase of the film thickness and volume fraction of void. In fact, RMS increased from 39 nm for 3:1 to 46 nm for 3:2 films in the 20 × 20-μm2 area. These findings imply that Br incorporation into FAPbIxBr3−x perovskite films promotes the densification of perovskite network, resulting in the increase in the free volume as a void and film thickness. In addition, prominent sub-gap absorption tail in the 3:2 film compared to that of 3:1 film suggests that the defect originates from the increased grain boundary of FAPbIxBr3−x (3:2) perovskite phase for larger Br composition ratio.
3.4. Role of IPA solvent in the synthesis of FAPbIxBr3−x perovskites
Figure 8a and b shows XRD pattern and FTIR spectra of a-PbIxBr2−x film before and after immersion into IPA solvent alone for 10 s. The inset of Figure 8a also shows AFM images of the corresponding PbIxBr2−x films. Despite immersion for only 10 s, the peak at 2θ = 12.7° attributed to the crystalline PbIxBr2−x (1 0 0) diffraction was observed, which suggests that the crystallization of the film can be attributed mainly to immersion in the IPA solvent. The FTIR spectra also revealed that the removal of the C═O-bending mode and C─H-stretching mode attributed to residual CHP in a-PbIxBr2−x was promoted by the IPA solvent. Thus, film crystallization of a-PbIxBr2−x was promoted by the removal of residual CHP in a-PbIxBr2−x and was accompanied by the formation of mesoporous crystallized PbIxBr2−x with large amounts of voids and/or grain boundaries.
Figure 9 shows n and k spectra for a-PbIxBr2−x before and after immersion into IPA solvent alone for 10 s. The best-fit optical models are shown in the upper panels. A prominent feature was that the n value decreased markedly with a red shift of the absorption edge onset corresponding to film crystallization with an increase in fvoid. The significant decrease in n is due to crystallization and the destruction of the mesoporous in a-PbIxBr2−x. These results imply that IPA had a significant effect to promote crystallization of the a-PbIxBr2−x film with an associated decrease in the bandgap due to a phase transition from 0D to 3D [Pb(IxBr1−x)6]4− octahedral clusters. These factors are the main contributors to the formation of the FAPbIxBr3−x perovskite network. Notable spectral features were the appearance of broad and sharp absorptions at 2.2 (probably due to defects/disorder) and 2.48 eV (A4+ → A4) assigned as the first exciton region in 2D PbIxBr2−x, in addition to those at 3.1, 3.5, and 3.9 eV due to film crystallization. In addition, no significant change in the film thickness was observed after immersion into the IPA solvent, but only the formation of voids, which suggests that immersion in IPA did not contribute to an increase of the film thickness.
From these results, the growth mechanism for the formation of FAPbIxBr3−x perovskites from an a-PbIxBr2−x layer and FAIxBr1−x/IPA solution is considered to be as follows (Figure 10) . FAPbIxBr3−x perovskites include an octahedral PbX6 (X = I, Br) complex with coordinated CHP solvent molecules. When a-PbIxBr2−x including ─CHP groups is immersed in the IPA solvent, C═O and C─H groups in coordinated CHP are preferentially removed by IPA solvent, which results in the promotion of a-PbIxBr2−x film crystallization and the formation of a large amount of voids and/or grain boundaries in the network. The diffusion of FAIxBr1−x into the voids/grain boundaries of the crystalline PbIxBr2−x phase is simultaneously promoted and leads to the formation of the FAPbIxBr3−x perovskite structure with an increase in the film thickness. In addition, no XRD diffraction peaks attributed to the perovskite structure were observed for vacuum-evaporated FAIxBr1−x as a precursor layer without a polar solvent such as IPA and CHP on a-PbIxBr2−x films, which implies that the IPA solvent promotes the growth of the perovskite structure.
The effect of Br incorporation into FAPbI3 perovskite films is to promote the densification of perovskite network, resulting in the increase in the free volume as a void and film thickness. In addition, prominent sub-gap absorption tail in the 3:2 film compared to that of 3:1 film suggests that the defect originates from the increased grain boundary of perovskite phase for larger Br compositional ratio.
The role of IPA solvent in the synthesis of FAPbIxBr3−x perovskites from an a-PbIxBr2−x thin layer and a solution of FAIxBr1−x in IPA solvent was investigated using SE combined with characterization by XRD and FTIR analysis. IPA played a significant role to promote crystallization of the a-PbIxBr2−x film through the removal of coordinated CPH from the a-PbIxBr2−x network, accompanied by the formation of grain boundaries, voids, and surface roughness. The diffusion of FAIxBr1−x into the voids and/or grain boundaries resulted in the simultaneous formation of large amounts of free volume and an increase in the film thickness, resulting in a decrease in the refractive index. The red shift of absorption edge from 3.4 to 1.73 eV is attributed to a phase transition from 0D to 3D [PbX6]4− octahedral clusters. These factors are the main contributors to the formation of the FAPbIxBr3−x perovskite network.
The author thanks Dr. Yohei Numata and Professor Tsutomu Miyasaka of Toin Yokohama University for fabrication of the perovskite thin film solar cells.