Fabrication and Characterization of Perovskite Solar Cells Using Metal Phthalocyanines and Naphthalocyanines ^†

Abstract

Fabrication and characterization based on experimental results for methylammonium lead iodide (MAPbI₃) perovskite solar cells using chemical-substituted metal phthalocyanines (MPc) and naphthalocyanines (MNc) as hole-transport materials have been performed to improve conversion efficiency (η) and stability. The purpose of this study was to fabricate and characterize a MAPbI₃ perovskite solar cell using t-butyl MPc and MNc as a hole-transporting layer to improve the photovoltaic performance and stability of η. Photovoltaic characteristics, morphology, crystallinity, and electronic structures were characterized in perovskite solar cells using MPc and MNc. The photovoltaic performance of the perovskite solar cell using t-butyl nickel phthalocyanine (NiPc) reached the maximum value of η at 13.4%. Incorporation of NiPc passivated the surface morphology by increasing the crystal grain size and supporting the carrier diffusion while suppressing carrier recombination near the grain boundary in the perovskite layer. Simulation using a SCAPS-1D program predicted the photovoltaic characteristics of the perovskite solar cell using NiPc. The photovoltaic mechanism was discussed on the basis of an energy diagram of the perovskite solar cell. The insertion of NiPc optimized energy levels near the highest occupied molecular orbital of NiPc and the valence band state of MAPbI₃, supporting a charge transfer related to short-circuit current density and η.

1. Introduction

Perovskite solar cells using metal phthalocyanines (MPc) and naphthalocyanines (MNc) have been fabricated and characterized to improve photovoltaic properties and stability [1,2,3,4]. Molecular-modified MPc and MNc, instead of conventional hole-transporting materials such as Spiro-OMeTAD, have the advantage of using hole-transporting materials in perovskite solar cells. These devices have excellent photovoltaic characteristics, which maintain over 20% in conversion efficiency under atmosphere [5,6]. MPc and MNc molecular structures, such as conjugated aromatic structure, central metal, and chemical substitution, reflect photovoltaic performances enhanced by semi-conductive properties with wide ranges of optical absorption in the ultraviolet–visible–near infrared region and thermal stability [7,8]. Photovoltaic properties of the lead halide perovskite solar cell are based on the state of carrier diffusion and charge transfer from the highest occupied molecular orbital (HOMO) of MPc to the valence band (VB) state of the perovskite crystal, and the band gap is related to the absorption wavelength [9].

In addition, chemical substitutions on MPc and MNc have a passivation effect on the defect and surface morphology near grain boundaries, supporting crystal growth and orientation while inhibiting decomposition in the perovskite layer [10,11,12,13,14]. These photovoltaic characteristics have been improved while suppressing carrier recombination in the perovskite layer with the passivation of modified MPc and MNc. The photovoltaic performance of conversion efficiency and stability of perovskite solar cells using t-butyl MPc and t-butyl MNc have been reported [1,10,12,15,16]. To investigate the effect of modified MPc and MNc on the photovoltaic mechanism, experimental results, such as photovoltaic and optical properties, surface morphology, and crystal structure, were analyzed by theoretical prediction using first principle calculation and molecular dynamic calculation [12,15,16]. Incorporation of the substituted MPc and MNc improved the photovoltaic performance of conversion efficiency exceeding 20% and stability [1,8,17,18,19]. The best efficiency achieved was 24.95% in a Cs_0.04Rb_0.02MA_0.02FA_0.92PbI₃ perovskite solar cell using methoxytriphenylamine CuPc [20]. These devices had thermal stability and long lifetimes, as compared with standard devices using Spiro-OMeTAD. Photovoltaic mechanisms have been discussed on the basis of energy diagram of the perovskite solar cell with one-dimensional solar cell simulation using a solar cell capacitance simulator (SCAPS-1D) program [21,22,23]. Photovoltaic performance based on the band gap, and carrier mobility was affected by defect density and layer thickness in the perovskite device.

The purpose of this study was to fabricate and characterize methylammonium lead iodine (MAPbI₃) perovskite solar cells using t-butyl MPc and MNc to improve conversion efficiency and stability. The effect of metal ions in MPc and MNc on the photovoltaic properties, surface morphology, crystallinity, orientation, and electronic structure was investigated. Photovoltaic characteristics, X-ray diffraction (XRD) patterns, and optical microscopy were measured. Electronic structures near the HOMO and lowest unoccupied molecular orbital (LUMO) of MPc and MNc were calculated. Simulation using a SCAPS-1D program predicted the photovoltaic properties and energy diagram of the perovskite solar cell using MPc. The photovoltaic mechanism is discussed using experimental results and simulation.

2. Experiments

A perovskite solar cell using MPc and MNc as hole-transporting materials was fabricated as reported previously in the literature [12,15,16]. Molecular structures of MPc and MNc (M = Ni²⁺, Zn²⁺, Pd²⁺, Cu²⁺) are shown in Figure 1. The standard fabrication was performed as reported previously in the literature [12,15,16]. TiO₂ precursor solutions were prepared, and the spin-coating process and annealing treatment were performed to form the compact and mesoporous TiO₂ layer on an F-doped tin oxide (FTO) substrate. The TiO₂ layer was ideal as a scaffolding substrate and electron-transporting layer for the perovskite solar cell. MAPbI₃ perovskite compounds were synthesized according to reaction Equation (1) as previously reported [12,15,16]. Methylammonium chloride (CH₃NH₃Cl) as a byproduct was decomposed and vaporized from the perovskite layer by heat treatment.

3 CH₃NH₃I + PbCl₂ → 2 CH₃NH₃Cl + CH₃NH₃PbI₃

(1)

The precursor solution of CH₃NH₃I (2.4 M), and PbCl₂ (0.8 M) with a desired mole ratio in N, N-dimethylformamide (0.5 mL), were introduced into the mesoporous TiO₂ by the spin-coating method at 1000 rpm for 5 s and 2000 rpm for 60 s using an air-blowing method. A solution of MPc was prepared as the hole-transporting layer [12,15,16]. The solution of MPc (Orient Chemical Industries, Osaka, Japan, 5 mg) was adjusted to a concentration of 10 mg/mL in chlorobenzene (0.5 mL). Samples of MPc and decaphenylcyclopentasilane (DPPS, OGSOL SI-30-15, Osaka Gas Chemicals, Osaka, Japan) in chlorobenzene were spin-coated at 2000 rpm for 60 s and annealed at 190 °C for 8 min. Solutions of Spiro-OMeTAD (Sigma-Aldrich, Tokyo, Japan) in chlorobenzene, lithium bis(trifluoromethylsulfonyl)imide (Li-TFSI, Tokyo Chemical Industry, Tokyo, Japan), tris(2-(1H-pyrazol-1-yl)-4-tert-butylpyridine) cobalt (III) tri[bis(trifluoromethane)sulfonimide] (FK209 Co (III) TFSI salt, Sigma-Aldrich, Tokyo, Japan) in acetonitrile, and 4-tert-butylpyridine were prepared and mixed as previously reported. The Spiro-OMeTAD solution was spin-coated on the DPPS layer at 4000 rpm for 30 s. All procedures were performed under an air atmosphere at 23 °C and a humidity of 36%. Finally, gold (Au) electrodes were evaporated as top electrodes using a metal mask for the patterning. Layered structures of the prepared photovoltaic cells were denoted as FTO/TiO₂/perovskite/MPc and MNc/DPPS/Spiro-OMeTAD/Au. Cells were prepared as no-sealed devices.

Prepared perovskite solar cells were stored in a drying desiccator at 22 °C and ~30% humidity. The J-V characteristics of these solar cells were measured under illumination at 100 mW cm⁻² with the use of an AM 1.5 solar simulator (San-ei Electric, XES-301S, Osaka, Japan) using a source measurement unit (Keysight, B2901A Precision SMU, Santa Rosa, CA, USA). Scans were performed in forward and reverse directions. These devices were characterized using three measurements of each cell, and average values were obtained from the averages of three measurements (η_ave). The external quantum efficiency (EQE, QE-R3011, Enli Technology, Osaka, Japan) of these cells was also measured using a source meter (Keithley Tektronix, 2450, Cleveland, OH, USA). The microstructures of cells were observed using an optical microscope (Eclipse E600, Nikon, Tokyo, Japan). Crystal structures were measured by X-ray diffraction (XRD, Bruker, D2 PHASER, Yokohama, Japan). Geometry optimization and energy calculation for MPc and MNc molecules were performed by the ab initio calculation using the density functional theory (DFT) method with the approximated wave functions of unrestricted B3LYP with LANL2DZ as the basis set (Gaussian 5.0, Gaussian Inc., Wallingford, CT, USA). Supercells of MAPbI₃ perovskite crystal as cubic system (a = 6.3910 Å) with a 2 × 2 × 2 crystal unit were constructed with the experimental results of the crystal structure obtained from the X-ray diffraction pattern [24]. MPc and MNc molecules were placed on the MAPbI₃ crystal, forming a hybrid composition while maintaining the intermolecular distance between the metal ion and iodine ion. Electron density distributions, energy levels at the LUMO and HOMO, the band gap (E_g), and total energies (E_total) of the hybrid compound were calculated. Simulation using a solar cell capacitance simulator (SCAPS-1D) program [25] was performed to characterize the photovoltaic properties of Jsc, Voc, FF, η, and EQE. Photovoltaic parameters were compared on the basis of experimental and theoretical results. A Mott–Schottky plot and energy diagram were simulated by a SCAPS-1D. The photovoltaic mechanism was discussed by an energy diagram of the perovskite solar cell using NiPc and ZnPc.

3. Results and Discussion

Photovoltaic characteristics of J-V curves of MAPbI₃ perovskite cells using MPc and MNc are shown in Figure 2a,b. The highest value of η in the reverse direction was obtained when hysteresis was generated in the forward and reverse directions of scanning. Photovoltaic parameters of Jsc, Voc, Rs, and Rsh are listed in

Table 1. Photovoltaic parameters of MAPbI₃ solar cells using MPc or MNc.

Devices	Jsc (mA cm−2)	Voc (V)	FF	Rs (Ω cm2)	Rsh (Ω cm2)	η (%)	ηave (%)
MAPbI3	19.5	0.802	0.664	3.67	7450	10.7	9.03
+NiPc	21.1	0.885	0.726	2.58	38,900	13.4	11.5
+ZnPc	15.3	0.760	0.535	3.48	327	6.71	4.70
+PdPc	13.1	0.615	0.430	3.28	209	3.47	2.88
+CuNc	19.1	0.873	0.689	4.45	7790	11.4	9.33
+ZnNc	16.3	0.840	0.586	3.09	306	8.03	5.64

. The addition of MPc and MNc affected photovoltaic parameters. In the case of NiPc and CuNc, the parameters of Jsc, Voc, FF, Rs, and Rsh related to η improved as compared with the parameters of MAPbI₃ cells using ZnNc, ZnPc, and PdPc. Photovoltaic parameters improved with increased amounts of NiPc. In the case of NiPc and CuNc at 10 mg per mL of solvent, conversion efficiency was achieved with maximum values of 13.4% and 11.4% as the best performances. In the case of NiPc at 18 and 25 mg/mL solvent, the photovoltaic parameter reduced, yielding values of η at 9.52% and 12.5%. Photovoltaic performance was based on the surface passivation on interfacial conditions.

From stability measurement, values of η for NiPc and CuNc gradually increased to about 13.6% during 115 days, and 12.3% during 62 days was the best performance. The stability of η in the solar cells was due to the promotion of crystal growth with the suppression of decomposition in the perovskite crystal. In the case of ZnPc, the value of η was obtained as about 6.7% and 7.4% during 33 days. These values of η were considerably low as compared with MAPbI₃ at η at 10.7% and 11.8% during 87 days. Recently, the characterization of perovskite solar cells using substituted phthalocyanine has been reported to improve the stability of the performance [1,12,15,16]. The best performance of the solar cell using t-butyl CuPc and t-butyl ZnPc achieved η of 18.8% and 15.5%, indicating long-term thermal stability based on crystallinity and surface coverage [10,26].

EQE curves of the solar cells using MPc are shown in Figure 2b. In the case of t-butyl NiPc, EQE increased to over 70% in the wavelength range of 400–550 nm. Photons were effectively converted to current in the ultra-visible region. Incorporation of NiPc supported carrier generation and diffusion while restraining carrier recombination in the perovskite layer. In the cases of ZnPc and PdPc, EQE decreased to less than 30% in the range of 400–750 nm. Additions of ZnPc and PdPc as passivation were not effective in inhibiting decomposition in the perovskite layer, reducing the Jsc related to EQE.

Surface morphologies of the perovskite layer with MPc were observed. Incorporation of NiPc improved the surface morphologies to a smooth structure with few areas of grain boundaries. The uniform layer supported carrier diffusion related to Jsc and η. The substituents of NiPc supported passivation near the defect site and the interfacing of the grain boundary in the perovskite layer. The photovoltaic performance remained during the aging time of 115 days. In the case of ZnPc, small black precipitation coagulated, forming rough and heterogeneous phases. Performance and stability were reduced. The trap densities of NiPc were measured to be about 0.93 × ${10}^{16}$ cm⁻³ by dark current–voltage characterization. Incorporating NiPc supported passivation, yielding increases in Jsc and η. The trap densities of ZnPc increased to 1.37 × ${10}^{16}$ cm⁻³, predicating a loss of photovoltaic performance.

Simulation using a SCAPS-1D program predicted the photovoltaic parameters of Jsc, Voc, FF, η, and EQE. Photovoltaic characteristics were considered on the basis of experimental and calculated results. Calculated and experimental results are shown in Figure 3 and

Table 2. Experimental and calculated parameters of solar cell using NiPc.

NiPc	Voc (V)	Jsc (mA cm−2)	FF	η (%)
Exp.	0.865	21.5	0.627	11.6
Calc.	1.092	22.9	0.463	11.6

and

Table 3. Photovoltaic parameters of solar cells using NiPc in SCAPS-1D simulation.

	FTO	TiO2	MAPbI3	NiPc	Spiro-OMeTAD
Thickness (nm)	250	400	500	60	100
Bandgap (eV)	3.50	3.2	1.55	1.70	2.9
Electron affinity (eV)	4.40	4.00	3.75	3.06	2.2
Dielectric permittivity (relative)	9.00	100	22.0	6.392	3.0
CB effective density of state (1/cm3)	2.2 × 1018	1.0 × 1020	3.1 × 1019	2.5 × 1020	2.5 × 1020
VB effective density of state (1/cm3)	1.8 × 1019	1.0 × 1020	3.1 × 1018	2.5 × 1020	2.5 × 1020
Electron thermal velocity (cm/s)	1.0 × 107	1.0 × 107	1.0 × 107	1.0 × 107	1.0 × 107
Hole thermal velocity (cm/s)	1.0 × 107	1.0 × 107	1.0 × 107	1.0 × 107	1.0 × 107
Electron mobility (cm/V s)	2.0 × 103	50	7.80	1.23 × 10−4	3.5 × 10−4
Hole mobility (cm/V s)	1.0 × 102	50	7.80	4.97 × 10−2	4.0 × 10−3
Donor density (1/cm3)	2.0 × 1019	1.0 × 1018	0	0	0
Acceptor density (1/cm3)	0	0	1.0 × 1016	1.0 × 1018	1.0 × 1019
Defect density (1/cm3)	1.0 × 1015	1.5 × 1017	9.3 × 1015	1.0 × 1015	1.0 × 1016

. As listed in Table 2, the calculated η was obtained to be about 11.6%, which was in agreement with the experimental results. However, the experimental parameters of Voc, Jsc, and EQE were slightly decreased as compared with the calculated results, as shown in Figure 3a,b and Table 2. The experimental behavior of the Jsc related to EQE originated from recombination as a carrier loss near the interface of the grain boundary in the perovskite layer. These photovoltaic properties were based on the morphologies, interfacial conditions, and defect densities listed in Table 3. The calculated defect densities for NiPc and ZnPc were obtained to 1.0 × 10¹⁵ and 8.40 × 10¹⁶, which was close to the measured values. Their photovoltaic properties depended on the internal structure.

Electronic density distribution, energy levels at the LUMO and HOMO, and the E_g of MPc were calculated. Electronic density distributions were similarly dispersed on the aromatic ring in NiPc, ZnPc, and PdPc. In the case of NiPc, energy levels at the LUMO and HOMO, and the E_g were obtained to be −2.84 eV, −5.15 eV, and 2.30 eV, as compared with the experimental results of the E_g at 1.54 eV. The calculated results of the E_g were slightly large as compared with the experimental results. The molecular interactions between MPc and MAPbI₃ affected energy levels at the LUMO and HOMO, and the E_g. In the case of MAPbI₃ with NiPc, energy levels at the LUMO and HOMO, the E_g, and total energy were obtained to be −3.53 eV, −5.23 eV, 1.70 eV, and −3539 eV. The calculated result of the E_g was close to the experimental results. The molecular interaction supported to form the stable crystal. The molecular interaction based on the hydrogen bond between the t-butyl group in MPc and the iodine ion worked near the interface. The t-butyl group was inserted into defects at the A-site of the crystal, and was easily stabilized as predicted by molecular dynamics calculation [12].

X-ray diffraction patterns were measured in the cells using MPc. In the case of NiPc, the lattice constant, crystallite size, and diffraction peak ratio of I₁₀₀/I₂₁₀ were obtained to be 6.274 Å, 451.7 Å, and 3.54. In the case of ZnPc, the lattice constant, crystallite size, and diffraction peak ratio of I₁₀₀/I₂₁₀ were obtained to be 6.274 Å, 506.4 Å, and 3.30, as compared with values of 6.274 Å, 429.9 Å, and 3.21 in the MAPbI₃ crystal. Incorporation of NiPc and ZnPc supported the formation of crystal of increased size. The crystallinities remained unchanged during 120 days. In the case of NiPc, uniform and smooth surface morphologies were observed. The uniform surface with low, uneven morphologies supported carrier diffusion by reducing carrier recombination. In the case of ZnPc, it was observed that heterogeneous phases with roughness formed, reducing the performances of Jsc, Voc, η, and EQE.

A Mott–Schottky plot is shown in Figure 4a. This device has a high built-in potential (Vbi) of 1.0 V. This result indicates a strong built-in electric field on the onset of space–charge collection in the device with the addition of NiPc. The photovoltaic mechanism was discussed using an energy diagram of perovskite solar cells. Simulation using a SCAPS-1D program predicted the photovoltaic characteristics of perovskite solar cells using NiPc. The energy diagram of the perovskite solar cell is shown in Figure 4b. Energy levels at the HOMO and LUMO of NiPc were close to the VB and conduction band states of MAPbI₃. Photovoltaic performances were based on the charge transfer from the VB state of MAPbI₃ to the HOMO of NiPc, supporting carrier diffusion related to Jsc and η. The hole-transporting layer using NiPc and Spiro-OMeTAD acted as a blocking layer to protect reverse current, reducing current loss. Energy levels at the HOMO and LUMO, and the E_g were optimized by molecular interactions between NiPc and MAPbI₃. The incorporation of NiPc had a passivation effect on the morphologies and interface of the grain boundary in the perovskite layer, as explained in previous papers [12,15,16]. The surface modification with the addition of NiPc improved the carrier transfer related to Jsc, EQE, and stability. In the case of ZnPc, photovoltaic characteristics were considered on the basis of experimental and calculated results, as shown in Figure 5a and

Table 4. Experimental and calculated parameters of solar cell using ZnPc.

ZnPc	Voc (V)	Jsc (mA cm−2)	FF	η (%)
Exp.	0.760	15.3	0.535	6.20
Calc.	0.795	15.6	0.510	6.30

. The calculated η was obtained to be about 6.30%, which agreed with the experimental results. The performance was based on the charge transfer from the VB state of MAPbI₃ to the HOMO of ZnPc while causing carrier recombination as carrier loss in the perovskite layer with defect density, as shown in Figure 5b and

Table 5. Photovoltaic parameters of solar cells using ZnPc in SCAPS-1D simulation.

	MAPbI3	ZnPc
Bandgap (eV)	1.55	2.20
Electron Affinity (eV)	3.75	3.06
Dielectric permittivity	22.0	6.39
CB effective density of state (cm3)	3.1 × 1018	2.7 × 1027
VB effective density of state (cm3)	3.1 × 1018	2.7 × 1027
Electron mobility (cm2/V)	7.80	1.23 × 10−4
Hole mobility (cm2/V)	7.40	1.10 × 10−2
Donor density (1/cm3)	1.0 × 1015	0
Acceptor density (1/cm3)	1.0 × 1015	1.0 × 1018
Defect density (1/cm3)	1.43 × 1018	8.40 × 1016

. The defect density in the MAPbI₃ layer using ZnPc increased by three orders of magnitude more than that of the layer using NiPc did. Photovoltaic performances were based on morphologies, internal structure, and defect density.

4. Conclusions

MAPbI₃ perovskite solar cells using t-butyl MPc and t-butyl MNc were fabricated and characterized to improve their photovoltaic properties and stability. Their photovoltaic properties, morphology, and crystallinity were measured and analyzed. In the case of NiPc, photovoltaic performance reached the maximum value of η at 13.4%. The surface passivation of NiPc supported crystal growth and orientation, improving carrier diffusion while suppressing carrier recombination near the grain boundary in the perovskite layer. Simulation using a SCAPS-1D program predicted photovoltaic properties. The photovoltaic mechanism was discussed using an energy diagram of the perovskite solar cell using NiPc. Energy levels near the HOMO and the E_g of NiPc were adjusted, supporting a charge transfer from the VB state of MAPbI₃ to the HOMO of NiPc. The incorporation of NiPc had a passivate effect on the surface morphology and crystallinity in the perovskite layer, improving the carrier transfer related to Jsc and η and stability.