Controlled nucleation in methylamine-treated perovskite films by artificial seeding and phase-field simulations

TL;DR

This work addresses parasitic nucleation during methylamine-assisted, seeded 2D crystallization of perovskite films. It combines phase-field simulations with an analytical framework to predict how seed spacing and the bare-substrate nucleation density $\eta$ govern unwanted nucleation, introducing a dimensionless criterion involving $D\sqrt{\eta}$. The authors validate the models across three material–substrate systems, showing that the only material-specific input needed is $\eta$, and that PF predictions align with experimental observations. The results offer a general, scalable strategy for designing seeded, large-grain 2D crystallization with broad applicability to optoelectronic devices.

Abstract

Large perovskite crystals with reduced defect density enable superior charge transport and stability. Therefore, controlling their nucleation and growth is key to advancing high-performance optoelectronic devices based on perovskite semiconductors. Millimeter-scale perovskite crystals can be synthesized as a continuous film through methylamine treatment, with nucleation sites directed by pre-patterned seeds. Nonetheless, certain configurations may lead to unwanted parasitic nucleation. To predict and mitigate this effect, we employ phase-field simulations alongside an analytical model. Their predictive capability is demonstrated across three distinct material-substrate systems, enabling precise control over nucleation and subsequent crystal growth. Notably, the only material-specific input required is the nucleation density (i.e., the number of crystals nucleated per unit area on an unpatterned substrate). This generality makes the models broadly applicable to diverse material systems for achieving controlled two-dimensional crystallization for improved optoelectronic device performance.

Figures (14)

• Figure 1: a) Schematic of the methylamine treatment process. b) Phase diagram of the methylamine process: For sufficiently high methylamine (MA) partial pressures $p_{MA}$ at sufficiently low temperatures $T$, the system enters a liquid intermediate state. Through a reduction in $p_{MA}$, the system is brought just across the phase boundary, where slow nucleation and recrystallization ensues. The quantitative values presented here were derived for triple cation perovskite on an ITO substrate. c) Images of a sample during recrystallization. d) Images of the successful artificial nucleation of the perovskite. Note that the underlying Au structure is not visible, only the resulting perovskite crystals. Where the nucleating structures are too far apart, additional 'parasitic' nuclei grow (an example is circled in blue).
• Figure 2: The average number of parasitic crystals in dependence on the pitch D. The shaded area depicts the standard deviation over 10 simulation runs. Note that the only difference between the simulations, for the same pitch, is the random fluctuations. Insets: Snapshots of the time evolution of a substrate without (left) or with (right) initial seed crystals in the simulation. The amorphous phase is depicted in blue, and the crystalline phase is shown in red.
• Figure 3: The number of parasitic crystals obtained from the simulation and calculated analytically (\ref{['equ:AnaModel']}). Inset: sketch visualizing the parameters of the analytical model. The three crystals (orange) represent an unit cell of the hexagonal lattice. The crystals grow with a growth speed of $v_g$, while additional crystals nucleate in the amorphous phase with a rate of $\kappa$. The arrows and shaded circles visualize the lower and upper integration limits of the model (see main text).
• Figure 4: Different experimental stacks a), b), c), and directly below the experimental results during recrystallization on unpatterned substrates d), e), f). The different material systems result in different nucleation densities $\eta$.
• Figure 5: Experimental results of the number of excess grains over $D\sqrt{\eta}$ for the three material systems, along with the simulation data and analytical model curves. The experimental results match the theoretical data. The numbers shown here are the mean values derived from multiple growth processes and image analyses. The error bars correspond to the standard deviation of these values. The values result from the analysis of between 6 and 15 image sequences per data point, where each image series tracks one recrystallization process. For more details, reference the methods.