Chemical Analysis of the Interface between Hybrid Organic-Inorganic Perovskite and Atomic Layer Deposited Al2O3

Abstract

Ultrathin metal oxides prepared by atomic layer deposition (ALD) have gained utmost attention as moisture and thermal stress barrier layers in perovskite solar cells (PSCs). We have recently shown that 10 cycles of ALD Al₂O₃ deposited directly on top of the CH₃NH₃PbI_{3- x}Cl _x perovskite material, are effective in delivering a superior PSC performance with 18% efficiency (compared to 15% of the Al₂O₃-free cell) with a long-term humidity-stability of more than 60 days. Motivated by these results, the present contribution focuses on the chemical modification which the CH₃NH₃PbI_{3- x}Cl _x perovskite undergoes upon growth of ALD Al₂O₃. Specifically, we combine in situ Infrared (IR) spectroscopy studies during film growth, together with X-ray photoelectron spectroscopy (XPS) analysis of the ALD Al₂O₃/perovskite interface. The IR-active signature of the NH₃⁺ stretching mode of the perovskite undergoes minimal changes upon exposure to ALD cycles, suggesting no diffusion of ALD precursor and co-reactant (Al(CH₃)₃ and H₂O) into the bulk of the perovskite. However, by analyzing the difference between the IR spectra associated with the Al₂O₃ coated perovskite and the pristine perovskite, respectively, changes occurring at the surface of perovskite are monitored. The abstraction of either NH₃ or CH₃NH₂ from the perovskite surface is observed as deduced by the development of negative N-H bands associated with its stretching and bending modes. The IR investigations are corroborated by XPS study, confirming the abstraction of CH₃NH₂ from the perovskite surface, whereas no oxidation of its inorganic framework is observed within the ALD window process investigated in this work. In parallel, the growth of ALD Al₂O₃ on perovskite is witnessed by the appearance of characteristic IR-active Al-O-Al phonon and (OH)-Al═O stretching modes. Based on the IR and XPS investigations, a plausible growth mechanism of ALD Al₂O₃ on top of perovskite is presented.

Keywords: Al2O3; X-ray photoelectron spectroscopy; atomic layer deposition; infrared spectroscopy; perovskite.

Figure 1

IR spectrum of the pristine ∼300 nm CH₃NH₃PbI_3–xCl_x perovskite.

Figure 2

Change in the N–H stretching mode of the ∼300 nm thick perovskite upon continuous exposure to cycles of ALD Al₂O₃.

Figure 3

XRD spectra of the perovskite film before and after deposition of 200 cycles of ALD Al₂O₃.

Figure 4

(a) Differential IR spectra (difference in absorbance of the perovskite with and without Al₂O₃) with increasing number of ALD cycles. (b) Integrated area of the N–H stretching mode in (a) as a function of number of Al₂O₃ cycles. Absorbance of the N–H stretching mode, which is negative compared to the baseline, increases in magnitude with increasing number of ALD cycles. The dashed line is a fit to the exponential function exp(-rx), where, x = number of Al₂O₃ cycles and r = effusion rate of NH₃ or CH₃NH₂ species per cycle.

Figure 5

Surface XPS spectra of (a) N_1s, (b) C_1s, (c) Pb_4f, and (d) I_3d peaks before and after deposition of 25 cycles of ALD Al₂O₃ (corresponding to <1 nm) on top of the perovskite film. Open circles, solid lines and dashed lines are measured data, peak fits, and cumulative fits, respectively.

Figure 6

Normalized integrated areas of (a) N_1s, Pb_4f, I_3d, and (b) Al_2p as a function of number of Al₂O₃ cycles on perovskite. The dashed lines show predictions assuming a layer-by-layer growth mechanism.

Figure 7

Surface XPS spectra of N_1s and I_3d peaks of the perovskite film before and after (a), (c) 200 TMA doses, and (b), (d) 200 H₂O vapor doses. Open circles and solid lines are measured data and peak fits, respectively.

Figure 8

Proposed reaction mechanism between TMA and perovskite.

Figure 9

(a) Thickness of ALD Al₂O₃ on top of Si with increasing number of ALD cycles derived from XPS and SE (b) Thickness of ALD Al₂O₃ on top of perovskite as a function of number of ALD cycles. The dashed line is a linear fit. The thickness calculation was repeated for three batches to calculate the standard deviation, and thereby the respective error bars. (c), (d) High angle annular dark field (HAADF) scanning TEM cross-sectional images of 200 cycles (14 ± 1 nm) ALD Al₂O₃ deposited on top of the perovskite film taken at different resolutions. (e) Corresponding overlapping EDX elemental mapping image.