Material Exchange Property of Organo Lead Halide Perovskite with Hole-Transporting Materials

Abstract

Using X-ray diffraction (XRD), it was confirmed that the deposition of hole-transporting materials (HTM) on a CH₃NH₃PbI₃ perovskite layer changed the CH₃NH₃PbI₃ perovskite crystal, which was due to the material exchanging phenomena between the CH₃NH₃PbI₃ perovskite and HTM layers. The solvent for HTM also changed the perovskite crystal. In order to suppress the crystal change, doping by chloride ion, bromide ion and 5-aminovaleric acid was attempted. However, the doping was unable to stabilize the perovskite crystal against HTM deposition. It can be concluded that the CH₃NH₃PbI₃ perovskite crystal is too soft and flexible to stabilize against HTM deposition.

1. Introduction

After the finding of the organo lead halide perovskite solar cell by the groups of Miyasaka, Snaith, Park and Grätzel [1,2,3,4], it has received significant attention in the last five years with improvement of the conversion efficiency from 3% [1] to a certified value of 20.1% [5]. The organo lead halide perovskite solar cells had been thought to be a solid-state device. However, with the recent efforts of scientist, it has emerged that the organo lead halide perovskite is an ion-conductive semiconductor [6,7]. Although the organo lead halide layer can give a clear perovskite crystal structure, as revealed by X-ray diffraction (XRD), it is just a moisture-sensitive and the most chemically-soluble salt.

Basically, printed devices should be deposited on a hard-soluble material to an easy-soluble one. Otherwise, the former material can be dissoluble into the latter one. For example, although CuSCN hole-transporting material (HTM) can work on the CH₃NH₃PbI₃ perovskite layer to be a solar cell [8], the CuSCN and CH₃NH₃PbI₃ perovskite can diffuse into each other and make a short circuit [8,9]. In order to improve the conversion efficiency of perovskite solar cells, this phenomena is a critical issue. In this paper, we have disclosed the chemical effects on the CH₃NH₃PbI₃ perovskite layer, which has been investigated by X-ray diffraction (XRD). The materials deposited on the CH₃NH₃PbI₃ perovskite layer were solvents for HTM (chlorobenzene and propyl sulfide) and the HTM layer itself 2,2’,7,7’-Tetrakis-(N,N-di-p-methoxyphenylamine)-9,9’-spirobifluorene (spiro-OMeTAD with dopants and CuSCN). In order to enhance the strength of the CH₃NH₃PbI₃ perovskite crystal, dopants of halide (chloride and bromide) and 5-aminovaleic acid, which can make a two-dimensional (2D) perovskite layer to enhance crystallinity and stability [10,11,12], were also adopted for this study.

2. Experiments

A commercial TiO₂-nanocrystalline paste (the code name “PST-18NR”, d = 18 nm, JGC Catalysts and Chemicals Ltd., Kawasaki, Japan) was diluted to one-tenth with ethanol for spin-coating. Porous TiO₂ layers were fabricated by a spin-coating method (5000 rpm for 30 s) on glass substrates, which were washed by an ultrasonic water bath with detergent and a UV/O₃ system beforehand and annealed at 500 °C for 30 min.

By a one-step process, 4 types of organo lead halide perovskite crystals were fabricated using N,N-dimethylformamide (DMF): (1) CH₃NH₃PbI₃ (CH₃NH₃I:PbI₂ = 2:1, in DMF); (2) CH₃NH₃PbCl_xI_3-x (CH₃NH₃I:CH₃NH₃Cl:PbI₂ = 1:1:1, in DMF); (3) CH₃NH₃PbBr_xI_3-x (methylammonium iodide (MAI):PbI₂:PbBr₂ = 2:0.9:0.1, in DMF); and (4) (HOOC(CH₂)₄NH₃)_x(CH₃NH₃)_1-xPbI₃ (MAI:5-aminovaleric acid(5-AVAI):PbI₂ = 25:1:25, in DMF). The amount of Pb in DMF was 1.2 M, consistently. Each solution was stirred at 70 °C and then coated on the mesoporous TiO₂ film by spin-coating at 2000 rpm for 30 s (5-s acceleration), followed by annealing at 100 °C for (1) to (3) and at 50 °C for (4).

For the two-step process CH₃NH₃PbI₃ perovskite deposition, 1.2 M of PbI₂ was dissolved in 2 mL of N,N-dimethylformamide by stirring at 70 °C, deposited on the mesoporous TiO₂ film by spin-coating at 6500 rpm for 20 s and dried at 100 °C for 30 min. Then, the TiO₂/PbI₂ film was dipped into a solution of CH₃NH₃I in 2-propanol (10 mg/mL) for 20 s and then rinsed with 2-propanol and spun to dry at 4000 rpm for 8 s (2-s acceleration). Finally, the deposited CH₃NH₃PbI₃ film was dried on a hotplate at 70 °C for 30 min.

A CuSCN hole conductor was deposited on CH₃NH₃PbI₃ layers by the doctor blading method at 65 °C (the CuSCN solution was prepared by dissolving 6 mg of CuSCN (Kishida Chemical Co. Ltd., Osaka, Japan) into 1 mL of propyl sulfide) [6,7,11]. 2,2’,7,7’-Tetrakis-(N,N-di-p-methoxyphenylamine)-9,9’-spirobifluorene (spiro-OMeTAD, Merck Japan, Tokyo, Japan) was deposited as an organic hole conducting material (HTM) with doping materials (4-tert-butylpyridine (TBP), lithium bis(trifluoromethylsulfonyl)imide (LiTFSI) and tris(2-(1H-pyrazol-1-yl)-4-tert-butylpyridine)cobalt(III) tris(bis(trifluoromethylsulfonyl) imide) in chlorobenzene [13]. The organic HTM solution was deposited using the spin-coating method at 4000 rpm. The crystal structures were characterized using X-ray diffraction (Miniflex II, Rigaku, Tokyo, Japan) with Cu Kα radiation.

3. Results and Discussions

At first, the solvent-rinsing effect on CH₃NH₃PbI₃ perovskite crystal variation was considered before the HTM deposition itself. Figure 1 shows the XRD patterns of the CH₃NH₃PbI₃ perovskite crystal after spin-coating chlorobenzene (solvent of spiro-OMeTAD) and propyl sulfide (solvent of CuSCN) on one-step processed CH₃NH₃PbI₃ perovskite layers. The original CH₃NH₃PbI₃ perovskite crystal showed the XRD peak at 13.7° and 468 counts. Although the intensity of XRD counts was related to the measurement setup, the height between the samples can be compared to discuss the variation of crystal structures using the same system consistently. It was surprising that just spin-coating chlorobenzene, which has been used for the spiro-OMeTAD deposition as an inert solvent against the CH₃NH₃PbI₃ perovskite crystal, can shift the peak to a higher angle and reduce the height of the XRD peak to half of the original one. The XRD peak shift to a higher angle means that the crystal lattice became shorter. This phenomenon may be due to the extraction of remaining DMF in the perovskite layer to chlorobenzene, which was explained as the “antisolvent method” published by Xiao et al. [14]. This result informs us that the remaining DMF in the perovskite crystal after the annealing procedure can be removed, because of the softness of the CH₃NH₃PbI₃ perovskite crystal [6,7]. In contrast, propyl sulfide (the solvent of CuSCN) did not change the XRD peak of the CH₃NH₃PbI₃ perovskite crystal, but reduced the XRD peak height. Since solvents (chlorobenzene and propyl sulfide) reduce the XRD peak of the CH₃NH₃PbI₃ perovskite crystal, the deposition of HTM by spin-coating can reduce the crystallinity of CH₃NH₃PbI₃ perovskite due to the solvent effect.

Figure 1. XRD patterns of one-step-processed CH₃NH₃PbI₃ perovskite layers varied by solvent rinsing on the perovskite layer.

Figure 1. XRD patterns of one-step-processed CH₃NH₃PbI₃ perovskite layers varied by solvent rinsing on the perovskite layer.

Figure 2 shows the XRD patterns of one-step-processed CH₃NH₃PbI₃ perovskite crystals with HTM layers. The peak at 15.8° represents the crystal of the CuSCN layer. It was surprising that CuSCN deposition enhanced the crystallinity of CH₃NH₃PbI₃ perovskite, which may be due to the Cu⁺-ion and SCN⁻-ion migration into CH₃NH₃PbI₃ perovskite, resulting in the enhancement of the perovskite crystal. Actually, CH₃NH₃Pb(SCN)₂I can also be a perovskite solar cell, effectively [15]. In contrast, the XRD pattern of CH₃NH₃PbI₃ perovskite did not change with spiro-OMeTAD deposition. Hence, it was considered that the solvent of spiro-OMeTAD (chlorobenzene) can shift the XRD peak position to the higher angle, but that the dopant (which may be lithium ion and TBP) can shift it back to the original position, due to the special phenomena of material migration in the CH₃NH₃PbI₃ perovskite crystal [6,7].

Figure 2. XRD patterns of one-step-processed CH₃NH₃PbI₃ perovskite layers varied by hole-conducting materials (HTM) on the perovskite layer. 2,2’,7,7’-Tetrakis-(N,N-di-p-methoxyphenylamine)-9,9’-spirobifluorene is represented as “spiro-OMeTAD”.

Figure 2. XRD patterns of one-step-processed CH₃NH₃PbI₃ perovskite layers varied by hole-conducting materials (HTM) on the perovskite layer. 2,2’,7,7’-Tetrakis-(N,N-di-p-methoxyphenylamine)-9,9’-spirobifluorene is represented as “spiro-OMeTAD”.

Figure 3 shows the XRD patterns of the two-step deposition of the CH₃NH₃PbI₃ perovskite crystal with the HTM layer. Due to the procedure of the two-step sequential deposition of the PbI₂ crystal into methylammonium iodide solution, some part of the PbI₂ can remain [9]. Although the CuSCN XRD peak can remain on the one-step deposited CH₃NH₃PbI₃ (15.8°, in Figure 2), it was found that the peak of CuSCN vanished on the two-step-deposited CH₃NH₃PbI₃ perovskite layer, which may be due to the attack of the excess amount of iodide ion from the PbI₂ layer on the CuSCN crystal. It was also proposed that the thiocyanate ion (SCN⁻) can migrate to the PbI₂ layer and can be exchanged with the ion with iodide ion (I⁻) to form the CH₃NH₃Pb(SCN)₂I perovskite crystal, resulting in the emission of the iodide ion to the CuSCN layer to be some amorphous phase. Hence, we can observe the shift of the PbI₂ layer. Moreover, the CuSCN can shift the XRD pattern of CH₃NH₃PbI₃ perovskite to a higher angle and lower intensity. Therefore, CuSCN can be mixed with the CH₃NH₃PbI₃ perovskite crystal effectively. In contrast, the effect of spiro-OMeTAD coating on the two-step-processed perovskite is also quite significant, because the spiro-OMeTAD coating causes the peak of PbI₂ to vanish and can shift the perovskite peak to the same position with the one-step-deposited CH₃NH₃PbI₃ perovskite. This PbI₂ decomposition may be due to the attack of the lithium ion and TBP on the PbI₂ crystal.

It was confirmed that the CH₃NH₃PbI₃ perovskite crystal was quite soft, and it was easy to change the lattice with the deposition of the HTM layer. In order to stabilize the CH₃NH₃PbI₃ perovskite crystal, we have attempted doping with other halogens (chloride and bromide) and 5-aminovaleric acid (AVA) in the CH₃NH₃PbI₃ perovskite crystal.

Figure 3. XRD patterns of the two-step-processed CH₃NH₃PbI₃ perovskite layers varied by hole-conducting materials (HTM) on the perovskite layer. 2,2’,7,7’-Tetrakis-(N,N-di-p-methoxyphenylamine)-9,9’-spirobifluorene is represented as “spiro-OMeTAD”.

Figure 3. XRD patterns of the two-step-processed CH₃NH₃PbI₃ perovskite layers varied by hole-conducting materials (HTM) on the perovskite layer. 2,2’,7,7’-Tetrakis-(N,N-di-p-methoxyphenylamine)-9,9’-spirobifluorene is represented as “spiro-OMeTAD”.

Figure 4 shows the doping effects of Cl⁻ and Br⁻ ions on the CH₃NH₃PbI₃ perovskite and HTM layers. Although there are some publication about Cl⁻ and Br⁻ ions, which can enhance the crystallinity and stability of the CH₃NH₃PbI₃ perovskite crystal [10,11], Figure 4 suggests that Cl⁻ and Br⁻ ions cannot stabilize the CH₃NH₃PbI₃ crystal against HTM deposition. It was confirmed that the doping of Cl⁻ and Br⁻ ions can greatly reduce or cause the XRD peak of CuSCN to vanish, which should arise around 15.8° (as in Figure 2). Actually, the chloride ion can improve the crystallinity of CH₃NH₃PbI₃ perovskite with over 10,000 counts of XRD (Figure 4a). However, the HTM deposition of spiro-OMeTAD or CuSCN can reduce the XRD intensity and shift the position of the XRD peak. Specially, the deposition of spiro-OMeTAD can reduce the XRD peak of the CH₃NH₃PbI₃ perovskite crystal very significantly. With the bromide ion, the XRD peak of CH₃NH₃PbI₃ perovskite can improve with the incorporation of CuSCN (Figure 4b). The vanishing effect of the CuSCN peak by the chloride ion is stronger than that by the bromide ion.

Figure 5 shows the XRD patterns of the 5-AVA-doped CH₃NH₃PbI₃ perovskite crystal by spin-coating deposition. Although only one XRD peak of 5-AVA-doped CH₃NH₃PbI₃ perovskite was observed below 10° in the literature [12], three intense peaks from 6° to 10° were observed as the 2D CH₃NH₃PbI₃ perovskite crystal. The difference from the published results may be due to the fabrication methods: spin-coating and drop casting. Without HTM, the peaks of the three-dimensional (3D) CH₃NH₃PbI₃ perovskite crystal (at 13.9°) and PbI₂ (at 12.9°) were quite small. Adding CuSCN HTM, the peak position of the 2D CH₃NH₃PbI₃ perovskite crystal did not change, but the peak height was decreased to less than half. At the same time, the peaks of the 3D CH₃NH₃PbI₃ perovskite crystal (at 13.9°) and PbI₂ (at 12.5°) were increased significantly. The peak height of the 3D CH₃NH₃PbI₃ perovskite crystal became very close to the one-step-processed CH₃NH₃PbI₃ perovskite. Adding spiro-OMeTAD, in contrast, the peak position of the 2D CH₃NH₃PbI₃ perovskite crystal was shifted to the lower angle, and the peak height was increased. Although the peaks of the 3D CH₃NH₃PbI₃ perovskite crystal and PbI₂ were increased, the increment by spiro-OMeTAD was half of that by CuSCN. Hence, spiro-OMeTAD did not change the 2D CH₃NH₃PbI₃ perovskite crystal, but some of the contents (maybe the lithium ion and TBP) can be inserted between the 2D CH₃NH₃PbI₃ perovskite layers, which can be explained by the peak shift of 2D perovskite to the lower angle. Such a 2D perovskite layer has stability against moisture [12,16]. However, it was confirmed that such a 2D perovskite also can be changed by HTM deposition.

Figure 4. XRD patterns of one-step-processed CH₃NH₃PbI₃ perovskite layers with chloride and/or bromide doping, varied by hole-conducting materials (HTM) on the perovskite layer with chloride doping (a) and with bromide doping (b). 2,2’,7,7’-Tetrakis-(N,N-di-p-methoxyphenylamine)-9,9’-spirobifluorene is represented as “spiro-OMeTAD”.

Figure 4. XRD patterns of one-step-processed CH₃NH₃PbI₃ perovskite layers with chloride and/or bromide doping, varied by hole-conducting materials (HTM) on the perovskite layer with chloride doping (a) and with bromide doping (b). 2,2’,7,7’-Tetrakis-(N,N-di-p-methoxyphenylamine)-9,9’-spirobifluorene is represented as “spiro-OMeTAD”.

Figure 5. XRD patterns of the one-step-processed CH₃NH₃PbI₃ perovskite layers with 5-aminovaleric acid (5-AVA) doping, varied by hole-conducting materials (HTM) on the perovskite layer. 2,2’,7,7’-Tetrakis-(N,N-di-p-methoxyphenylamine)-9,9’-spirobifluorene is represented as “spiro-OMeTAD”.

Figure 5. XRD patterns of the one-step-processed CH₃NH₃PbI₃ perovskite layers with 5-aminovaleric acid (5-AVA) doping, varied by hole-conducting materials (HTM) on the perovskite layer. 2,2’,7,7’-Tetrakis-(N,N-di-p-methoxyphenylamine)-9,9’-spirobifluorene is represented as “spiro-OMeTAD”.

However, it was found that the resulting PbI₂ layer from the 2D CH₃NH₃PbI₃ perovskite layer has higher stability than the 3D perovskite crystal. Figure 6 shows a picture of PbI₂ layers after water deposition (with different pH) of CH₃NH₃PbI₃ perovskite layers (with/without 5-AVA doping). The CH₃NH₃PbI₃ perovskite layers with/without 5-AVA doping have been decomposed to PbI₂ by dipping in water. When the amount of eluted Pb in the water was measured, no elution of Pb was observed using neutral water (pH = 7). However, alkaline water solution (pH = 12 of NaOH aq. solution) dissolved the CH₃NH₃PbI₃ perovskite layers completely. The observed amounts of Pb ion using CH₃NH₃PbI₃ perovskite layers with/without 5-AVA doping were 800 mg/m² and 725 mg/m², respectively. This small variation may be due to the difference of thickness after the 5-AVA doping. The significant difference of doping was confirmed using acidic water solution (pH = 2, HCl aq. solution). The 5-AVA doping can enhance the stability of the PbI₂ crystal against acidic water solution. This kind of effort to suppress the leakage of Pb elution should be improved for the application of organo lead halide perovskite solar cells.

Figure 6. Pictures of the one-step-processed CH₃NH₃PbI₃ perovskite layers with/without 5-aminovaleric acid (5AVA) doping, dipped in water, aqueous HCl solution (pH = 2) and aqueous NaOH solution (pH = 12). The numbers inserted in the picture show the amount of the lead dissolved into the water, the aqueous HCl solution (pH = 2) and the aqueous NaOH solution (pH = 12).

Figure 6. Pictures of the one-step-processed CH₃NH₃PbI₃ perovskite layers with/without 5-aminovaleric acid (5AVA) doping, dipped in water, aqueous HCl solution (pH = 2) and aqueous NaOH solution (pH = 12). The numbers inserted in the picture show the amount of the lead dissolved into the water, the aqueous HCl solution (pH = 2) and the aqueous NaOH solution (pH = 12).

Figure 7 shows a summary of the reaction image between the CH₃NH₃PbI₃ perovskite crystal for the deposition of solvents (chlorobenzene and propyl sulfide) and HTM (spiro-OMeTAD and CuSCN). Chlorobenzene can extract DMF in the non-dried perovskite layer [14]. Moreover, the chlorobenzene may have a similar effect on the annealed perovskite layer (from Figure 1). In contrast, propyl sulfide did not shift the peak position, but decreased the intensity due to the decomposing of the smaller crystal (from Figure 1). The thiocyanate ion (SCN⁻) can be exchanged by the iodide ion (I⁻) in PbI₂ [15] and by the chloride (Cl⁻) and bromide (Br⁻) ions doped in perovskite crystals to be CH₃NH₃Pb(SCN)₂I perovskite [15], resulting in the prevention of the CuSCN crystal on the CH₃NH₃PbI₃ perovskite layer (from Figure 2, Figure 3 and Figure 4). The Cu⁺ ion and the SCN⁻ ion can enhance the crystallinity of CH₃NH₃PbI₃ perovskite. However, the chloride ion and the excess amount of I⁻ (from PbI₂) can suppress the enhancement of the CH₃NH₃PbI₃ perovskite crystallinity. For the 2D (5-AVA)_x(CH₃NH₃)_1-xPbI₃ perovskite crystal, the lithium ion and TBP in spiro-OMeTAD HTM can be inserted between the layered perovskite crystal (as in Figure 7b, from Figure 5). The deposition of CuSCN HTM can convert the 2D crystal to a 3D one.

In order to discuss the CH₃NH₃PbI₃ perovskite crystal transformation, the water content (purity) in the solvents, HTM materials and atmosphere should be explained, because water can decompose the CH₃NH₃PbI₃ perovskite crystal into PbI₂ ([16]; as in Figure 6). In this work, the humidity of the laboratory was 60% to 80%. The contents of water in the HTM materials and the solvents were not measured. However, the HTM depositions did not enhance the PbI₂ peak during the experiments, basically (only one exception was the two-step-processed CH₃NH₃PbI₃ perovskite with CuSCN HTM in Figure 3). Therefore, it is considered that the effect of water can be negligible. For the further study of the stability of perovskite solar cells, the amount of water in HTM would be very important.

Figure 7. Summary of the reaction scheme of organo lead halide perovskite layers for CH₃NH₃PbX₃ (X: I, Cl or Br) (a) and for (HOOC(CH₂)₄NH₃)_x(CH₃NH₃)_1-xPbI₃ (b).

Figure 7. Summary of the reaction scheme of organo lead halide perovskite layers for CH₃NH₃PbX₃ (X: I, Cl or Br) (a) and for (HOOC(CH₂)₄NH₃)_x(CH₃NH₃)_1-xPbI₃ (b).

Moreover, some molecules (DMF, DMSO, etc.) were known to be inserted into the crystal lattice, as studied by Yang et al. [5]. In this work, the experiments had been performed in laboratory fume hoods. No glovebox was used for the experiments. Hence, it was considered that the effect of volatile solvent on the perovskite surface from air was eliminated effectively. Of course, there is a chance that the changes by solvent exposure are irreversible or returned (recovered) after annealing of the perovskite layer, as reported by Jeon et al. [17], which is generally carried out for the completion of a stand-alone perovskite layer. However, the aim of this paper is to show the instability against HTM deposition, which will not be annealed after the HTM deposition. Moreover, Cu²⁺, SCN⁻ and Li⁺ are non-volatile ions. Hence, we would like to skip the annealing effect of crystal recovery after HTM deposition for future studies.

The decomposition mechanism of the CH₃NH₃PbI₃ perovskite crystal by the lithium ion and TBP was speculated as the exchanging effect with the methyl ammonium ion (CH₃NH₃⁺). Basically, TBP is a strong base, which can exchange H⁺ with water or CH₃NH₃⁺. In contrast, it can be speculated that the lithium ion and TBP can be inserted into the hydrogen bonding (R-COOH···HOOC-R; the dot line “···” represents hydrogen bonding) in the gap between the 5-AVA-ion (HOOC(CH₂)₄NH₃⁺)-doped 2D perovskite crystals. Further research will be performed to elucidate the exact mechanism.

4. Conclusion

In conclusion, the effect of HTM deposition on the CH₃NH₃PbI₃ perovskite crystal was systematically investigated by XRD. It was confirmed that although inorganic HTM can work effectively as organic HTM, CuSCN can react with CH₃NH₃PbI₃. Therefore, porous TiO₂ is necessary to block short circuits [9]. In order to achieve concrete results, further experiments should be done, for example TEM, XPS, SEM, Auger spectroscopy, X-ray absorption fine structure (XASF), and so on. However, it is quite important to show the preliminary results and to investigate the inside of the CH₃NH₃PbI₃ perovskite crystal. It can be concluded that the CH₃NH₃PbI₃ perovskite crystal is too soft and flexible to be stabilized against HTM deposition. In order to enhance the perovskite-crystal crystallinity and stability, the mixed halogen effect and 2D perovskite have been published [10,11,12]. However, no method for them can enhance the crystal stability. For the further application of CH₃NH₃PbI₃ perovskite solar cells, the migration of HTM materials to the perovskite layer should be considered, because such post-deposition of HTM can decompose the soft perovskite crystal. Therefore, the perovskite crystal should be deposited as the last material deposited in the perovskite solar cells. For example, the structure of full-printed solar cells as, <FTO/compact-TiO₂/porous-TiO₂/porous-ZrO₂/(porous-NiO)/porous-carbon (perovskite was inserted into the porous electrodes)> is quite promising, because the perovskite crystal can be formed at the end of the process [12,18].