Near-band-edge optical responses of solution-processed organic–inorganic hybrid perovskite CH₃NH₃PbI₃ on mesoporous TiO₂ electrodes

Sensitized solar cells (SSCs) have attracted considerable attention as a promising light-energy conversion device because of their low fabrication cost, process simplicity, and robust nature.^1,2) A standard SSC comprises a transparent conducting film, a porous TiO₂ electrode, a light-harvesting sensitizer, and a hole-transporting material (HTM).^1–3) Photogenerated electrons are promptly injected into TiO₂, and the carrier separation at the sensitizer/TiO₂ interface produces a photovoltage in the SSCs. To realize efficient SSC devices, a ruthenium-based dye and an iodine-based electrolyte are widely used as a sensitizer and an HTM, respectively.^2,3) Recently, a power conversion efficiency of more than 11% has been realized in this type of SSC.⁴⁾ An all-solid-state SSC has also been developed;^5,6) however, its power conversion efficiency was much lower than that of a liquid-state SSC.

A breakthrough was realized with the use of organic–inorganic hybrid perovskite CH₃NH₃PbX₃ (X = Cl, Br, I) as a sensitizer.^7–14) In this lead-halide-perovskite-based SSC, spiro-OMeTAD [2,2',7,7'-tetrakis-(N,N-di-p-methoxyphenylamine)9,9'-spirobifluorene] is typically used as a solid-state HTM. The power conversion efficiency of such a device is being rapidly improved, and presently, it stands at more than 15%.¹²⁾ Surprisingly, a high power conversion efficiency was found to be obtained by the use of mesoporous Al₂O₃ instead of TiO₂; in this case, photogenerated electrons in CH₃NH₃PbX₃ cannot be injected into Al₂O₃ owing to the large energy difference between the conduction bands of Al₂O₃ and CH₃NH₃PbX₃.^9,10) Porous Al₂O₃ is considered to only serve as a scaffold on which TiO₂ is coated.^9,10) This suggests that the light-energy conversion mechanism of a perovskite-based solar cell is completely different from that of a conventional SSC, and therefore, it is called a mesoscopic heterojunction solar cell. To further realize the potential of this new type of SSC, it is essential to clarify the fundamental optical properties of a CH₃NH₃PbX₃-based solar cell. The band-gap energy of CH₃NH₃PbI₃ on mesoporous TiO₂ was reported to be 1.5 eV as estimated by diffuse reflectance (DR) spectroscopy. However, the photoluminescence (PL) peak appearing at 1.6 eV is much above this band-gap energy.⁹⁾ The above-band-gap PL is not usually observed in conventional semiconductors. These results strongly suggest that even the band-gap energy, the most important parameter of a semiconductor, is not determined for this cell. As such, the optical properties of solution-processed CH₃NH₃PbI₃ on mesoporous TiO₂ near the band gap remain unclear. Better understanding of its near-band-edge optical responses is important for the improvement of solar cell performance.

In this work, we studied the near-band-edge optical responses of solution-processed CH₃NH₃PbI₃ on mesoporous TiO₂ electrodes by means of DR, photoconductance (PC), PL, and transient absorption (TA) spectroscopy. At room temperature, the DR spectrum shows that the extrapolated line's zero-crossing point is 1.59 eV. On the other hand, PL shows a broad peak at 1.6 eV, indicating that the band-gap energy exceeds 1.6 eV. The PL excitation (PLE) spectrum also shows a broad band at 1.64 eV. The TA spectrum shows a negative peak at 1.61 eV. We measured the temperature dependence of the DR and PL spectra. The steepness of the absorption edge is linearly dependent on the temperature at high temperatures. Therefore, we conclude that the Urbach tail determines the room-temperature band-edge optical absorption, and the band-gap energy is ∼1.61 eV at room temperature.

For sample fabrication, CH₃NH₃I was prepared by the slow addition of a solution of methylamine in methanol (40%, Wako Pure Chemical Industries, 65.0 mL, 568 mmol) into an aqueous solution of hydroiodic acid (57 wt %, Wako Pure Chemical Industries, 60.4 mL, 592 mmol) in a 500 mL round-bottom flask at 0 °C over 10 min, followed by stirring for 2 h. The white precipitates were collected by filtration and purified by recrystallization from a mixed solvent of diethyl ether and methanol. After filtration, the obtained crystals were dried at 60 °C under vacuum for 24 h to give 75.6 g (476 mmol, 84%) of CH₃NH₃I. The porous TiO₂ layer was deposited on the glass substrate by spin coating at 5000 rpm for 30 s using a TiO₂ paste (JGC Catalysts and Chemicals PST-18NR) diluted in ethanol (1 : 2.5, weight ratio). After drying at 100 °C, the TiO₂ films were gradually heated to 550 °C over 1 h, baked at this temperature for 30 min, and cooled to room temperature. The porous TiO₂ films were infiltrated with PbI₂ (99.999%, Sigma-Aldrich) by spin-coating of a PbI₂ solution in N,N-dimethylformamide (460 mg/mL, Wako Pure Chemical Industries Super Dehydrated) at 6500 rpm for 5 s in an argon-filled glove box. After keeping at 70 °C and drying for 30 min, the films were dipped in a solution of CH₃NH₃I in 2-propanol (10 mg/mL, Wako Pure Chemical Industries Super Dehydrated) for 20 s and rinsed with 2-propanol.¹¹⁾ As shown in the right inset of Fig. 1, we confirmed that the solar cell using the fabricated CH₃NH₃PbI₃ works well. (See Ref. 12 for the solar-cell structure.) It showed high open-circuit voltage (V_OC = 0.96 V), short-circuit current (J_SC = 17.2 mA/cm²), and fill-factor (FF = 0.62). The power-conversion efficiency reached 10.2%.

Zoom In Zoom Out Reset image size

To study the optical absorption spectrum, we utilized DR spectroscopy because of the large light scattering due to mesoporous TiO₂ electrodes. The acquired DR spectrum was converted to the Kubelka–Munk function F(R), which is approximately proportional to the absorption coefficient, according to the relation F(R) = (1 − R)²/(2R), where R is the diffuse reflectivity. The PC spectrum was obtained using an SMO-250III (Bunkoukeiki). PL and PLE measurements were performed using a Si CCD camera with a monochromator. A wavelength-tunable Ti:sapphire (continuous-wave) laser was used as an excitation light source. The TA spectrum was measured using a standard pump–probe technique in which a femtosecond white-light pulse was used as a probe light. The pump pulse energy was 1.91 eV.

Figure 1 shows the (a) DR and PC spectra, (b) PL and PLE spectra, and (c) TA spectrum at room temperature. The DR and PC spectra are similar to each other and show a significant increase above 1.6 eV, which is attributed to a band-to-band optical transition. The left inset shows the Tauc plot of the DR spectrum for a direct-gap semiconductor, i.e., [F(R)hν]^γ plotted as a function of the excitation photon energy. Here, we adopt γ = 2 for direct-gap semiconductors without excitonic effects.^15,16) The dotted line indicates the linear extrapolation. The zero-crossing point is 1.59 eV, which is consistent with the previously reported band-gap energy.^9,10) It should be noted that the zero-crossing point derived in the Tauc region¹⁷⁾ (interband region, typically $\alpha \gtrsim 10^{4}$ cm⁻¹) is generally regarded as a band-gap energy while it remains unclear where the observed DR spectrum corresponds to the Tauc region in our sample.

As shown in Fig. 1(b), the PL spectrum has a single Gaussian peak at 1.60 eV. Because the emitted photon energy is usually below the band-gap energy under weak photoexcitation, this result indicates that the band-gap energy exceeds 1.60 eV. This is inconsistent with the zero-crossing energy of 1.59 eV in the DR spectrum. The PLE spectrum is also plotted in the same figure; it shows a broad peak at 1.64 eV. It is considered that the band-gap energy is located below the PLE peak. Therefore, we can speculate from PL and PLE spectroscopy that the band-gap energy lies in the range of 1.60–1.64 eV. Note that the decrease in PLE in the high-energy region might be caused by the large nonradiative recombination rate in the near-surface region, which reduces the PL efficiency under high energy excitation, because of the short penetration depth of excitation light.¹⁸⁾

The TA spectrum immediately after excitation (5 ps) also shows a negative peak at 1.61 eV, as shown in Fig. 1(c). The negative TA intensity indicates photobleaching, which usually appears in the neighborhood of the band-gap energy of direct-gap semiconductors. This result strongly suggests that CH₃NH₃PbI₃ is a direct-gap semiconductor. The TA peak energy of 1.61 eV is consistent with the PL and PLE peak energies. On the basis of near-band-edge PL, PLE, and TA spectra, we concluded that the band-gap energy of CH₃NH₃PbI₃ on mesoporous TiO₂ electrodes is 1.61 eV at room temperature.

To account for the discrepancy between the estimated band-gap energy (1.61 eV) and zero-crossing energy (1.59 eV) of the DR spectrum, we consider two possibilities: (i) the 1.60 eV PL component does not originate from CH₃NH₃PbI₃ but from other compounds unintentionally mixed in the sample, and (ii) the absorption edge of the DR spectrum does not reflect the band-gap energy owing to, for instance, a band-tail state such as an Urbach tail below the band edge. To examine these two possibilities, we studied the temperature dependences of the PL and DR spectra.

Figure 2 shows the PL and DR spectra at different temperatures. The PL peak energy and DR onset energy show good agreement at all temperatures. In the high-temperature region (150–300 K), the PL peak and DR onset energy are located at ∼1.6 eV and shift to the low-energy side with a decrease in temperature, indicating a reduction in band-gap energy. Below 150 K, a two-step variation of the DR and two PL peaks are observed. This is attributable to the tetragonal-to-orthorhombic structural phase transition, the transition temperature of which is reported to be 160 K.¹⁹⁾ Two different phases coexist at around the phase transition temperature.

Zoom In Zoom Out Reset image size

Moreover, the steepness of the absorption edge in the DR spectra increases with a decrease in temperature. Such temperature-dependent steepness of the absorption edge is not observed in ideal direct-gap semiconductors. A similar temperature-dependent spectrum below the band-gap energy is known as an Urbach tail, for which the absorption coefficient is described by α ∝ exp(σ(E − E_g)/k_BT).²⁰⁾ Here, σ, E, and E_g are the steepness coefficient, photon energy, and band-gap energy, respectively. For a quantitative estimation of the steepness and onset energy of the absorption edge, we calculated the first derivative of F(R), dF/dE. We plotted the dF/dE spectra at 80, 140, 200, and 300 K in Fig. 3(a). The data near the absorption edge can be fitted by one or two Gaussian functions. Two peaks of the dF/dE spectrum below 140 K clearly indicate a two-phase coexistence. The spectral width, which corresponds to the steepness of the absorption edge, decreases with temperature. Above 200 K, the spectral width linearly depends on the temperature, as summarized in the inset of Fig. 3(b). This result suggests that the high-temperature absorption edge is an Urbach tail. In this case, it is difficult to estimate the band-gap energy of CH₃NH₃PbI₃ from the DR spectrum shown in the inset of Fig. 1 at high temperature. We conclude that the Urbach tail below the band edge causes the complicated near-band-edge optical responses.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In Fig. 3(b), we summarize the PL peak energy and DR onset energy as functions of temperature. Here, we defined the DR onset energy as the peak energy of dF/dE. The temperature dependences of the PL and DR peak energies are similar to each other. At 150 K, both DR and PL show large variations owing to the tetragonal-to-orthorhombic phase transition, as mentioned above. This temperature dependence means that the observed PL originates from CH₃NH₃PbI₃ and not from the other compounds, and therefore, we can determine the band-gap energy of CH₃NH₃PbI₃ on the basis of the PL and PLE peak energies (shown in Fig. 1). The optical measurements conducted in this study show that the band-gap energy of solution-processed CH₃NH₃PbI₃ on mesoporous TiO₂ electrodes is ∼1.61 eV at room temperature.

In conclusion, we studied the optical properties of CH₃NH₃PbI₃ on mesoporous TiO₂ electrodes by means of DR, PC, PL, PLE, and TA. On the basis of the temperature-dependent optical spectra, we concluded that the discrepancy of the absorption edge in DR and PL is attributable to the Urbach tail. We determined the band-gap energy of CH₃NH₃PbI₃ on mesoporous TiO₂ electrodes. Our fundamental finding should help provide deep insights into the photoconversion mechanism of CH₃NH₃PbI₃-based solar cells.

Acknowledgments