High-performance photovoltaic perovskite layers fabricated through intramolecular exchange

See allHide authors and affiliations

Science  12 Jun 2015:
Vol. 348, Issue 6240, pp. 1234-1237
DOI: 10.1126/science.aaa9272

Taking in more sun

Most efforts to grow superior films of organic-inorganic perovskites for solar cells have focused on methylammonium lead iodide (MAPbI3). However, formamidinium lead iodide (FAPbI3) has a broader solar absorption spectrum that could ultimately lead to better performance. Yang et al. grew high-quality FAPbI3 films by starting with a film of lead iodide and dimethylsulfoxide (DMSO) and then exchanging the DMSO with formamidinium iodide. Their best devices achieved power conversion efficiencies exceeding 20%.

Science, this issue p. 1234


The band gap of formamidinium lead iodide (FAPbI3) perovskites allows broader absorption of the solar spectrum relative to conventional methylammonium lead iodide (MAPbI3). Because the optoelectronic properties of perovskite films are closely related to film quality, deposition of dense and uniform films is crucial for fabricating high-performance perovskite solar cells (PSCs). We report an approach for depositing high-quality FAPbI3 films, involving FAPbI3 crystallization by the direct intramolecular exchange of dimethylsulfoxide (DMSO) molecules intercalated in PbI2 with formamidinium iodide. This process produces FAPbI3 films with (111)-preferred crystallographic orientation, large-grained dense microstructures, and flat surfaces without residual PbI2. Using films prepared by this technique, we fabricated FAPbI3-based PSCs with maximum power conversion efficiency greater than 20%.

The tremendous improvements in device architecture (13), high-quality film formation methodologies (46), and compositional engineering of perovskite materials (79) over the past 3 years have led to rapid improvements in the power conversion efficiency (PCE) of perovskite solar cells (PSCs). Although solar-to-electric PCEs of up to 18% have been reported for PSCs (10), developing technologies further to achieve PCEs near theoretical values (>30%) continues to be an important challenge in making the solar cell industry economically competitive.

Formamidinium lead iodide (FAPbI3) is a perovskite material that can potentially provide better performance than methylammonium lead iodide (MAPbI3) because of its broad absorption of the solar spectrum. In addition, FAPbI3 with the n-i-p architecture (the n-side is illuminated with solar radiation) exhibits negligible hysteresis with sweep direction during current-voltage measurements (813). However, it is more difficult to form stable perovskite phases and high-quality films with FAPbI3 than with MAPbI3. Various methodologies such as sequential deposition (4), solvent engineering (5), vapor-assisted deposition (14), additive-assisted deposition (15, 16), and vacuum evaporation (6) can now produce high-quality films of MAPbI3 with flat surfaces and complete surface coverage by controlling its rapid crystallization behavior and have led to substantial improvements in the PCE of MAPbI3-based PSCs.

Among these methodologies, two-step sequential deposition and solvent engineering are representative wet processes that can yield perovskite films for high-performance PSCs. In the sequential deposition process, a thin layer of PbI2 is deposited on the substrate; methylammonium iodide (MAI) or formamidinium iodide (FAI) is then applied to the predeposited PbI2 to enable conversion to the perovskite phase. This process involves crystal nucleation and growth of the perovskite phase because of solution-phase or solid-state reaction between PbI2 and an organic iodide such as MAI or FAI (4, 13, 17, 18). However, the sequential reaction of organic iodides with PbI2 that occurs from the surface to the inner crystalline regions of PbI2 has been ineffective in producing high-performance perovskite films that are >500 nm in thickness because of incomplete conversion of PbI2, peeling off of the perovskite film in solution, and uncontrolled surface roughness. In contrast, the solvent-engineering process uses the formation of intermediate phases to retard the rapid reaction between PbI2 and organic iodide in the solution. Although this process has been successfully used to form dense and uniform MAPbI3 layers, it has not been explored for FAPbI3 (5).

To deposit a uniform and dense FAPbI3 layer, Snaith et al. added a small amount of aqueous hydrogen iodide (HI) to a solution mixture containing PbI2, FAI, and dimethylformamide (DMF) (11). Very recently, Zhao et al. reported the deposition of highly uniform and fully covered FAPbI3 films using FAI and HPbI3, which is formed by the reaction of PbI2 and HI in DMF (19). The HI in the PbI2 layers retards the rapid reaction between FAI and PbI2. In addition, the release of HI from PbI2 at high temperatures allows the formation of a FAPbI3 layer by solid-state reaction with the neighboring FAI molecules. Stated differently, this process can be regarded as the transformation of PbI2-HI-FAI into FAPbI3, similar to the formation of MAPbI3 via the PbI2–dimethylsulfoxide (DMSO)–MAI phase in the solvent-engineering process (5).

However, we observed that the solvent-engineering process, which is effective for depositing dense and uniform MAPbI3 layers, yields FAPbI3 layers with pinholes and a rough surface. Although aspects of FAPbI3 film quality, including coverage and uniformity on the substrate, have been improved, the performance of FAPbI3 solar cells still lags behind that of MAPbI3-based PSCs (8), implying that more sophisticated deposition techniques are necessary for fabricating high-quality, thick FAPbI3 films (>500 nm) that would enable sufficient absorption up to a wavelength of 840 nm.

As expected from the conversion of PbI2(DMSO)-MAI to MAPbI3 (5), the DMSO molecules intercalated in PbI2 can be easily replaced by external FAIs because of their higher affinity toward PbI2 relative to DMSO; the FAI molecules experience ionic interactions, whereas DMSO participates in van der Waals interactions (5, 20). Highly uniform and dense predeposited PbI2-DMSO layers could be directly converted to FAPbI3 because the inorganic PbI2 framework would be retained. FAPbI3 crystallization by the intramolecular exchange process (IEP) of DMSO intercalated in PbI2 with FAI was schematically shown in Fig. 1A. The intramolecular exchange between DMSO and FAI can be described as Embedded Image (1)and does not induce volume expansion, unlike the FAPbI3 formed with FAI intercalating into pristine PbI2 (discussed below), because the molecular sizes of DMSO and FAI are similar.

Fig. 1 PbI2 complex formation and x-ray diffraction.

(A) Schematics of FAPbI3 perovskite crystallization involving the direct intramolecular exchange of DMSO molecules intercalated in PbI2 with formamidinium iodide (FAI). The DMSO molecules are intercalated between edge-sharing [PbI6] octahedral layers. (B) XRD patterns of (a) as-prepared PbI2(DMSO)2 powders, (b) vacuum-annealed PbI2(DMSO) powders, and (c) as-deposited film on fused quartz substrate using PbI2(DMSO) complex solution. (C) TGA of PbI2(DMSO)2 (red line) and PbI2(DMSO) (dark blue line). (D) XRD patterns of (a) as-formed film of FAPbI3 by IEP, and (b) FAPbI3 powder.

In this work, we report on the synthesis of a PbI2(DMSO) precursor with excellent capabilities for molecular exchange with FAI at low temperatures during the spinning process, as well as the fabrication of highly efficient FAPbI3-based PSCs with certified PCEs exceeding 20%. To synthesize the PbI2(DMSO) precursors, we obtained precipitates by pouring toluene as a nonsolvent into 1.0 M PbI2 solution dissolved in DMSO. The x-ray diffraction (XRD) pattern of the resulting complex [Fig. 1B(a)] matched that of the PbI2(DMSO)2 phase (5, 20). The as-prepared PbI2(DMSO)2 was then annealed at 60°C for 24 hours in vacuum to obtain PbI2(DMSO) by removal of 1 mol DMSO. The XRD pattern of the vacuum-annealed powder [Fig. 1B(b)] did not match that of PbI2(DMSO)2, implying that the PbI2(DMSO)2 transformed into a different phase by releasing some DMSO molecules. The content of DMSO in the as-annealed powder was estimated by thermogravimetric analysis (TGA). TGA was suitable for this purpose because the only volatile species in the powder was DMSO. The TGA results of the PbI2(DMSO)2 and PbI2(DMSO) complexes are shown in Fig. 1C. The PbI2(DMSO)2 complex exhibited a two-step decomposition process with weight loss of 12.6% at each step, whereas the vacuum-annealed PbI2(DMSO) complex showed a single-step decomposition. The decomposition of both the complexes was completed at the same temperature (138.6°C). The powders obtained by vacuum-annealing PbI2(DMSO)2 complex at 60°C can be regarded as one of the most thermodynamically stable forms among the various crystalline PbI2(DMSO)-based complexes, which are similar to those of PbBr2(DMSO) and PbCl2(DMSO) (21). The DMSO content of the vacuum-annealed PbI2(DMSO) complex was also checked by elemental analysis, which yielded H = 1.0% (1.1%); and C = 4.1% (4.4%), where the values expressed in parentheses indicate the theoretical mass percent for a given element in C2H6SOPbI2.

To fabricate FAPbI3-based PSCs through IEP between DMSO and FAI (MABr) using predeposited PbI2(DMSO) layers and a FAI (MABr) solution, we first confirmed that the PbI2(DMSO) phase was retained even after spin-coating with the PbI2(DMSO) precursor dissolved in DMF. The XRD pattern for film coated on a fused silica substrate was compared with that of the initial precursor. As seen in Fig. 1B(c), the XRD pattern for the as-coated film was consistent with that of the PbI2(DMSO) complex powder, although its crystallinity was lower. The as-coated PbI2(DMSO) film also had a flat and dense surface, as shown in the field emission scanning electron microscopy (FESEM) image in fig. S1 (22). Next, we investigated the formation of mixed FAPbI3/MAPbBr3 by IEP. We recently reported that the coexistence of MA/FA/I/Br in the PbI2 skeleton improved the phase stability of FAPbI3 (10). The formation of mixed FAPbI3/MAPbBr3 layers via IEP was controlled by coating the solution mixture with different weight ratios of MABr to FAI, dissolved in isopropyl alcohol, on the predeposited PbI2(DMSO) layers (see below). It is evident from Fig. 1D(a) that well-crystallized FAPbI3-based films were formed by IEP. The XRD pattern for the FAPbI3 film derived from the PbI2(DMSO) complex film exhibits dominant (–111) and (–222) diffraction peaks at 13.9° and 28.1°, respectively, corresponding to the FAPbI3 trigonal perovskite phase (P3m1), in contrast with the XRD patterns of the FAPbI3 powder [Fig. 1D(b)] (13). The intensity ratio of the (–123) peak at 31.5° to the (–222) peak at 28.1° was 0.05. This value was much smaller than the corresponding intensity ratios (0.8) for the FAPbI3 powder. Thus, IEP leads to high-quality pure FAPbI3–based films with preferred orientation along the [111] axis.

Figure S2 (22) presents the current density−voltage (J-V) curves measured under standard AM 1.5G (air mass 1.5 global) illumination, as well as the external quantum efficiency (EQE) spectra of the fabricated cells with FAPbI3-based layers fabricated with various amounts of MABr (0 to 20 wt%). The onset wavelength in the EQE spectra near 830 nm showed a nonlinear blue shift with increasing amounts of MABr, indicating that there is unsymmetrical competition between FAI and MABr in forming the FAPbI3-MAPbBr3 perovskite phase through an intramolecular exchange reaction. Nonetheless, the highest PCE of 19.2% was achieved for the film fabricated from a FAI solution containing 15 wt% MABr. To accurately determine the composition of the FAPbI3-based layer, we investigated the lattice parameter using XRD and the band gap using the EQE for the film showing the best performance. Figure S3 (22) shows the pseudocubic lattice parameter for (FAPbI3)1–x(MAPbBr3)x as a function of x, in which the composition was controlled by a previously reported method (10). In this study, the pseudocubic lattice parameter of the FAPbI3/MAPbBr3 film fabricated by IEP with a FAI solution containing 15 wt% MABr is 6.348 Å. As indicated in fig. S4 (22), the lattice parameter can be assigned as x = ~5, corresponding to (FAPbI3)0.95(MAPbBr3)0.05. This result is in agreement with the value estimated using the band gap (1.49 eV) from EQE [fig. S3 (22)], because pure FAPbI3 has a band gap of 1.47 eV and (FAPbI3)0.85(MAPbBr3)0.15 has a band gap of 1.55 eV (10). Fortunately, the simultaneous introduction of both MA+ cations and Br anions in FAPbI3 even after incorporating 5 mol% of MAPbBr3 serves to stabilize the perovskite phase (10).

After comparing the absorption coefficients of FAPbI3 and MAPbI3 at wavelengths beyond 800 nm, we noted that the thickness of a FAPbI3 layer needed to be higher than the optimal thickness of a typical perovskite layer with a band gap of ~1.55 eV (300 to 400 nm) to guarantee full light harvesting around 800 nm (21, 23). We deposited FAPbI3-based layers with thickness of ~500 nm, and fabricated devices consisting of fluorine-doped tin oxide (FTO)–glass/barrier layer (bl)-TiO2/mesoporous (mp)-TiO2/perovskite/polytriarylamine (PTAA)/Au (n-i-p architecture), as shown in the cross-sectional FESEM image of Fig. 2A. FESEM plane-view images of the device with film derived from PbI2(DMSO) complex and PbI2 films are shown in Fig. 2B. The FAPbI3 film derived from PbI2(DMSO) exhibited a dense and well-developed grain structure with larger grains than the FAPbI3 film derived from PbI2. Figure 2C(a) shows the J-V curves measured via reverse and forward bias sweep for one of the best-performing solar cells. The devices we fabricated also showed no hysteresis. Here, we believe that the hysteresis is highly dependent on the perovskite materials (FAPbI3 or MAPbI3) and cell architecture (n-i-p or p-i-n), although the ferroelectric properties of the perovskite itself are more likely to be the origin of the hysteresis in PSCs (24, 25). Thus, FAPbI3-based PSCs with n-i-p architecture show negligible hysteresis between the reverse and the forward scan in the I-V characteristics. In contrast, FAPbI3-based cells consisting of FTO/NiO/perovskite/PCBM/LiF/Al (p-i-n architecture) showed very strong hysteresis [fig. S4 (22)]. Values of short-circuit current density (JSC), open-circuit voltage (VOC), and fill factor (FF) determined from the J-V curves were 24.7 mA cm−2, 1.06 V, and 77.5%, respectively, and correspond to a PCE of 20.2% under standard AM 1.5G illumination. Figure 2C(b) shows the EQE spectrum and integrated JSC for one of the best-performing solar cell. The high JSC is attributed to a very broad EQE plateau of >85% in the illumination wavelength range of 400 to 780 nm and broad light-harvesting up to a long wavelength of 840 nm, owing to the relatively low band gap (1.47 eV) of FAPbI3. The JSC value (24.4 mA cm−2) obtained by integrating EQE spectrum agreed well with that derived from the J-V measurement. The PCE of the best-performing cell (20.2%) was certified by the standardized method in the PV calibration laboratory, which confirmed a PCE of 20.1% under AM 1.5 G full-Sun illuminations [fig. S5 (22)].

Fig. 2 SEM observations and J-V and EQE measurements.

(A) Cross-sectional FESEM image of the device consisting of FTO-glass/bl-TiO2/mp-TiO2/perovskite/PTAA/Au. (B) The comparison of FESEM surface images of FAPbI3-based layer formed on mp-TiO2 by IEP and conventional method. (C) (a) J-V curves of best device measured with a 40-ms scanning delay in reverse (from 1.2 V to 0 V) and forward (from 0 V to 1.2 V) modes under standard AM 1.5G illumination, and (b) EQE spectra for best device and integrated JSC.

To gain more insight into the enhanced performance of the FAPbI3-based PSCs, we compared the properties of the films fabricated by IEP with those obtained from a conventional sequential process. A sequential reaction such as interdiffusion between FAI/MAI and PbI2 through thermal annealing in organic iodide/PbI2 multilayer films has been used to form perovskite FAPbI3/MAPbI3 films from inorganic PbI2 films in the conventional process (17, 18). Thus, considerable volume expansion occurs in the sequential deposition process based on PbI2 because of the growth of perovskite crystals with the insertion of organic iodides into PbI2 skeleton (14, 23). As expected, an initial PbI2 film with thickness of ~290 nm was doubled to 570 nm for the film formed by the reaction of PbI2 with FAI [Table 1 and fig. S6 (22)].

Table 1 Comparison of layer thickness before and after FAPbI3 phase is formed by conventional and intramolecular exchange process (IEP).

The thin PbI2 and PbI2(DMSO) layers were deposited on a fused quartz glass, and their layer thickness was measured by alpha-step IQ surface profiler.

View this table:

In contrast, the change in thickness observed by the application of the FAI (MABr) solution to the predeposited PbI2(DMSO) film was negligible. In fact, the reaction between FAI (MABr) and PbI2(DMSO) was completed within 1 min during spin-coating and the FAPbI3 perovskite phase was formed without sequential annealing. However, in a conventional process using PbI2 films, annealing at high temperature is required to achieve interdiffusion. Figure 3A compares XRD patterns for as-formed and annealed films by IEP and conventional process from PbI2(DMSO) complex film and PbI2 film, respectively; there is no appreciable difference in XRD patterns between as-formed and annealed film. This result confirms that the FAPbI3-based layer is formed by the IEP of DMSO and FAI (MABr) without additional annealing process. In addition, such an exchange can considerably favor crystallization into perovskite, compared to conventional interdiffusion from PbI2, and led to an increase in the XRD peaks intensity after annealing at 150°C for 20 min. However, the as-formed film with PbI2 showed XRD patterns assigned to PbI2, FAI, and FAPbI3, and a (002) peak at 12.5° corresponding to the PbI2 still remained after annealing at same temperature and time with IEP. In particular, the FAPbI3 film prepared by IEP shows preferred orientation in the (111) direction compared to FAPbI3 film annealed after preparing it by conventional process.

Fig. 3 Comparison of x-ray diffractions, performance, and reproducibility between IEP and conventional process.

(A) XRD patterns of as-formed and annealed film for FAPbI3-based layers formed by IEP (red line) and conventional (blue line) process. α, #, and * denote the identified diffraction peaks corresponding to the FAPbI3 perovskite phase, PbI2, and FAI, respectively. (B) Representative J-V curves for FAPbI3-based cells fabricated by IEP and conventional process. (C) Histogram of solar cell efficiencies for each 66 FAPbI3-based cells fabricated by IEP and conventional process.

The advantages of IEP become further apparent upon comparing the J-V curves and PCEs of FAPbI3-based devices derived from PbI2(DMSO) complex films and conventional PbI2 films (Fig. 3, B and C). The devices based on FAPbI3 fabricated from PbI2(DMSO) showed superior PCEs with smaller deviations in the value, compared to those prepared from conventional PbI2 films. High-efficiency solar cells with average PCEs of >19% could be produced with a high degree of reproducibility by using the IEP. This study provides an effective protocol for fabricating efficient and cost-effective inorganic-organic hybrid heterojunction solar cells.

Supplementary Materials

References and Notes

  1. See supplementary materials on Science Online.
  2. Acknowledgments: Supported by the Global Research Laboratory Program, the Global Frontier R&D Program on Center for Multiscale Energy System funded by the National Research Foundation in Korea, and a grant from the KRICT 2020 Program for Future Technology of the Korea Research Institute of Chemical Technology and SKKU-KRICT DRC program.
View Abstract

Stay Connected to Science

Navigate This Article