Two-dimensional metal-halide perovskites (2D MHPs) have attracted considerable attention in recent years as promising semiconducting materials for photovoltaic, (1) optoelectronic, (2,3) and spintronic applications. (4) Compared to their three-dimensional (3D) counterparts, 2D MHPs consist of alternating organic and inorganic layers, which offer a high degree of tunability in the number of layers, their structural arrangement, and the choice of organic spacers, which directly influence their electronic properties such as the bandgap. (5) Furthermore, their quantum-well-like structures result in strong quantum and dielectric confinement, leading to large exciton binding energies and tightly bound excitons. (6,7) The unique crystal structure of 2D MHPs provides enhanced environmental stability, (8) strong photoluminescence and electroluminescence, (9,10) and anisotropic carrier transport characteristics. (11) In addition, the wide variety of available organic spacers enables flexible molecular design, allowing for fine-tuning of material properties and engineering of unique functionalities such as ferroelectricity and spin selectivity. (12−14) Recent research has focused extensively on understanding the fundamental properties of energy carriers and structural dynamics in these materials, inspiring a range of advanced optoelectronic applications. (15−17)

Compared to inorganic semiconducting materials such as Si and GaAs, MHPs exhibit notably soft lattice characteristics. (18) In addition, 2D MHPs typically have a more compliable lattice than their 3D counterparts, as evidenced by lower elastic moduli. (19) As a result, 2D MHPs are more susceptible to lattice distortions, intrinsic phase transitions, and structural changes in response to external perturbations, such as temperature variations or charge carrier excitation. Billing et al. synthesized a series of 2D MHPs and systematically investigated their steady-state phase transitions using differential scanning calorimetry and X-ray crystallography. (20,21) Their study revealed that the phase transition temperature can be tuned by adjusting the length of the organic spacer cations. Other research shows that temperature-induced phase transitions in 2D MHPs can be effectively regulated by the chemistry of the organic spacers. (22,23) Several recent studies have employed time-resolved methods to explore the interactions between photoexcited charge carriers and the soft lattices of 2D MHPs. Time-resolved X-ray diffraction experiments revealed a transient relaxation and ordering of the octahedral framework driven by photoexcited charge carriers. (24) Zhang et al. employed ultrafast electron diffraction to discover a light-induced, ultrafast reduction in antiferro-distortion of the octahedra, showing that the interaction between the electron–hole plasma and the lattice can be regulated by adjusting the rigidity of 2D MHPs. (25)

Figure 1a illustrates the crystal structures of the three 2D MHPs investigated in this work. All three materials are n = 1 lead iodide MHPs, each containing organic spacer cations with varying alkyl chain lengths. The compounds can be chemically denoted as [CH₃(CH₂)_N−1NH₃]₂PbI₄, where N = 4 corresponds to butylammonium lead iodide or (BA)₂PbI₄, N = 5 corresponds to pentylammonium lead iodide or (PA)₂PbI₄, and N = 8 corresponds to n-octylammonium lead iodide or (OA)₂PbI₄. Thin films of these materials were fabricated by hot-casting onto standard glass substrates. Optical microscopy images (Figure S1) reveal that the films exhibit large grain sizes ranging from microns (μm) to tens of μm. Based on the interference peaks in the reflectance spectra in the below-bandgap regime (Figure S2), (29) the thicknesses of the films are estimated to be around 1 μm. X-ray diffraction data (Figure 1b) confirm a high degree of parallel alignment of the octahedral layers relative to the substrate. Furthermore, photoluminescence (PL) spectra (Figure 1c) indicate the high crystalline quality of the films, as evidenced by the relatively narrow PL peaks centered at excitonic wavelengths near 500 nm for these n = 1 lead iodide MHPs.

All three compounds exhibit reversible, first-order structural phase transitions in response to temperature variations. (20,21) Specifically, (BA)₂PbI₄ undergoes a transition between two orthorhombic phases (both within the Pbca space group) over a temperature range of 223 to 293 K. (20) This phase transition is characterized by a notable hysteresis, as well as spectral shifts in the PL peak and exciton absorption onset, which are associated with changes in the electronic bandgap. (30,31) The low-temperature (LT) to high-temperature (HT) phase transition is driven by an order–disorder transition of the BA cations. This change unlocks large-amplitude, anharmonic tilting of the octahedral networks and results in a spectral broadening of the low-wavenumber Raman peaks. (30) In the case of (PA)₂PbI₄, three distinct phases have been observed, including two monoclinic phases (space group of P2₁/a) at 173 and 293 K, and an orthorhombic phase (space group of Pbca) at 333 K. (20) Our present study for (PA)₂PbI₄ focuses on the transition between the monoclinic phase at 293 K and orthorhombic phase at 333 K. For (OA)₂PbI₄, our measurements capture all three of its observed phases, (21) including a monoclinic phase (space group of P2₁/a) at low temperature, and two orthorhombic phases near and above room temperature (vide infra).

We first focus on our results for (BA)₂PbI₄, one of the most widely studied 2D MHPs in the literature. (5,32,33) Since the phase transition temperature of MHP thin films can be influenced by film morphology and substrate-induced strain, (34−36) we characterized the phase transition of our hot-cast (BA)₂PbI₄ film using steady-state optical reflectance spectroscopy at normal incidence. As shown in Figure 2a, the strong exciton resonance of (BA)₂PbI₄, centered at around 500 nm, is manifested as a prominent dip in the reflectance spectrum, accompanied by a peak on the longer wavelength side. (37,38) The reflectance peaks of the LT and HT phases are spectrally separated by approximately 20 nm. Temperature-dependent reflectance spectra reveal a phase transition temperature of approximately 269 K upon heating and 252 K during cooling, indicating a relatively significant hysteresis of 17 K. As illustrated in Figure 2a (bottom), the LT phase exhibits a lower absorption onset wavelength, giving it a yellower appearance compared to that of the HT phase.

Importantly, the reflectance maps in Figure 2b show a significant change in the reflectance near the exciton resonance during the phase change. This is further illustrated by the plot in Figure 2c, which depicts the reflectance at 515 nm, corresponding to the wavelength of the reflectance peak in the HT phase. Based on this observation, we hypothesized that an impulsive thermal excitation, with a sufficient instantaneous temperature rise to induce a phase transition (from the LT to the HT phase), should result in a strong and distinguishable relative change in reflectance, denoted as ΔR/R. In steady-state conditions, the ΔR/R resulting from a temperature variation can be expressed as $[R(T_1)?R(T_0)]/R(T_0)$, where T₁ and T₀ represent two different temperatures. Under this definition, Figure 2d presents three ΔR/R curves, calculated using reflectance data at four different temperatures: 255 and 265 K, where (BA)₂PbI₄ resides in the LT phase, and 275 and 285 K, corresponding to the HT phase. The curves of $\mathrm{\Delta }R/R_{255\to 265?\mathrm{K}}$ and $\mathrm{\Delta }R/R_{275\to 285?\mathrm{K}}$ capture the effects of lattice heating on the reflectance in the LT and HT phases, respectively. $\mathrm{\Delta }R/R_{255\to 265?\mathrm{K}}$ exhibits a derivative-like line shape skewed toward the negative side, while $\mathrm{\Delta }R/R_{275\to 285?\mathrm{K}}$ displays a prominent positive peak centered around 500 nm. When the exciton resonance of (BA)₂PbI₄ is modeled as a damped harmonic oscillator (i.e., using the Lorentzian model), the spectral shapes of ΔR/R arise from a combination of changes in the damping factor, oscillator strength, and resonance frequency. (39,40) Notably, a distinctively different line shape is observed for $\mathrm{\Delta }R/R_{265\to 275?\mathrm{K}}$ (going from the LT to HT phase), which features a strong and broad negative dip centered at around 495 nm, with positive lobes on either side, indicative of a prominent bandgap shift.

After establishing the steady-state reflectance properties of (BA)₂PbI₄ and observing the distinguishable effects of the phase transition on reflectance, we conducted time-resolved VPVP spectroscopy at varying temperatures. The pump, with a pulse width of ∼170 fs, was tuned to a wavelength of 3200 nm, which is resonant with the C–H and N–H bond stretching vibrations of the BA cations. The broadband probe, with a pulse width of ∼1 ns, was delayed by time intervals from a single nanosecond to tens of μs. The pump-induced change in probe reflectance is represented as $\mathrm{\Delta }R/R=[R(t)?R(0)]/R(0)$, where R(t) is the probe reflectance at delay time t after the pump, and R(0) is the reflectance without pump excitation. Figure 3a presents two transient spectral maps of ΔR/R, both obtained at 242 K but under different pump fluences. For the ΔR/R measured at a lower fluence of 6.6 mJ·cm^–2, the transient change in reflectance primarily consists of a derivative-like feature spanning the 470 to 490 nm range. When comparing this transient ΔR/R to the steady-state analogue obtained within the LT phase (Figure 2d), we find that the transient ΔR/R is largely associated with a pump-induced temporary increase in lattice temperature. Notably, ΔR/R rapidly reaches a maximum following pump excitation, which is due to the fast (nanosecond) thermal equilibration between the excited organic vibrations and the rest of the phonon bath in the material. (41) The subsequent decay in ΔR/R is a result of the temperature decrease as heat dissipates from the (BA)₂PbI₄ film into the glass substrate. Additionally, a second, much weaker ΔR/R feature is observed in the 495–515 nm range, which can be attributed to a minor fraction of the HT phase present in the film dominated by the LT phase.

Interestingly, with a slight increase in pump fluence from 6.6 to 8.0 mJ·cm^–2, the ΔR/R spectral map (Figure 3a, right) undergoes a significant qualitative change. First, the negative lobe of ΔR/R broadens, extending to nearly 510 nm on the red side. Additionally, unlike at lower fluence, the ΔR/R signal does not rise instantaneously after pump excitation; instead, it grows in amplitude over the first few μs, followed by a decay on the tens of μs time scale. To identify and distinguish different processes obtained at the 8.0 mJ·cm^–2 pump fluence, we performed a global principal component analysis of the transient spectral ΔR/R data. The analysis, as summarized in Figure 3b, reveals that two components are needed to adequately reproduce the ΔR/R data, each with a distinct time scale and spectral shape. The amplitude of each component as a function of probe wavelength, termed the decay-associated spectrum (DAS), is shown in Figure 3b (left), while their corresponding temporal dynamics are presented in Figure 3b (right).

By comparing the DAS in Figure 3b with the steady-state data-derived ΔR/R spectra shown in Figure 2d, we find that the first DAS component closely matches $\mathrm{\Delta }R/R_{255\to 265?\mathrm{K}}$, suggesting that it can be attributed to the heating effect of the LT phase. Notably, the second DAS component, which exhibits a symmetric spectral shape with a broad negative lobe at 495 nm and two positive bands on either side, corresponds well with the steady-state spectrum of $\mathrm{\Delta }R/R_{265\to 275?\mathrm{K}}$. Consequently, this second ΔR/R component can be assigned to a transient phase transition induced by vibrational pump excitation. Examining their temporal dynamics shown in Figure 3b (right), we observe that the first ΔR/R component rises within the first 1 μs, while the second ΔR/R component shows a rise within the first 2 μs. In other words, despite subnanosecond lattice temperature rise, the growth of the HT phase occurs relatively slowly compared to other prototypical phase-change materials, such as VO₂ and chalcogenides, (42,43) where large-scale lattice reorganization is involved in their phase transitions. On the other hand, the μs time for the phase transition of (BA)₂PbI₄ is still considered fast, requiring time-resolved optical spectroscopy for its observation. This is because the poor thermal conductivity of 2D MHPs likely limits the heat transfer rate to slower than μs, making our observation inaccessible with quasi-static experiments.

We then varied the pump fluences in the VPVP experiments. Figure 3c shows the kinetics of the ΔR/R principal component associated with the transient phase transition under three different fluences, focusing on the first 3 μs. In our VPVP experiments, the transient LT-to-HT phase transition is not limited by the rate of heat transfer (in contrast to the HT-to-LT phase transition, which is strongly influenced by the heat transfer rate). However, despite a nonlinear increase in the degree of transient phase transition with increasing pump fluence, the time constant for the transient growth of the HT phase remains nearly unchanged (i.e., growth of the HT phase always takes about 2 μs). The relatively weak fluence dependence of the time constant, as shown in Figure 3c, suggests that the rate of transient phase transition does not strongly depend on transient lattice overheating. One possible explanation for this is that the phase transition proceeds via a heterogeneous nucleation and growth mechanism, potentially occurring at grain and twin boundaries in the 2D MHP films. These boundaries can accommodate the strain induced by the phase transition, thereby lowering the activation energy. The nucleation and growth of the HT phase are inherently slow, as they involve changes in the ordering and hydrogen bonding of the organic cations, as well as the cooperative tilting of the octahedral layers. Additionally, the large hysteresis in the phase transition temperature suggests that the film may exhibit spatial variation in its phase transition temperature, potentially due to strain and defects in the films. (34,44) Consequently, higher pump fluences result in a larger fraction of the film experiencing sufficient overheating to undergo the phase transition.

Additional transient spectral maps of ΔR/R, acquired at different temperatures in the range of 240–257 K are shown in Figures S3 and S4. Although the steady-state LT-to-HT phase transition, as shown in Figure 2b, primarily occurs at 269 K, we found that the pump-induced transient phase transition can only be observed at lower temperatures, ranging from 240 to about 254 K. This corresponds to the low-temperature side of the hysteresis range for the phase transition. The reason for this is the repetitive nature of the VPVP experiments, which requires the (BA)₂PbI₄ film to revert to the LT phase after each pump-induced LT-to-HT phase transition. When the VPVP experimental temperature exceeds 255 K, the HT-to-LT phase transition becomes less likely, as confirmed by the steady-state data shown in Figure 2b (right), which in turn hinders the VPVP experiments.

By performing global analysis on the transient ΔR/R data acquired at different temperatures from 240 to 251 K (Figures S3 and S4), we obtained the temperature-dependent kinetics of the principal component associated with the pump-induced phase transition, as shown in Figure 3d. The data reveal a decreasing amplitude of the principal component, which we attribute to the reduced amount of untransitioned LT phase at higher temperatures, that undergo pump-induced phase transition. Additionally, the phase transition rate slightly decreases with increasing temperature, suggesting that those untransitioned LT-phase domains require a higher phase-transition activation energy. At even higher temperatures of 254 and 257 K, we find that the (BA)₂PbI₄ film primarily consists of the HT phase, even though these temperatures are significantly lower than the steady-state LT-to-HT phase transition temperature of 269 K. This is likely due to cumulative pump-induced LT-to-HT phase transitions occurring during the VPVP experiments.

We then performed similar VPVP experiments to investigate the phase transition kinetics of (PA)₂PbI₄ and (OA)₂PbI₄, with the transient ΔR/R spectral maps presented in Figure 4a–c. For (OA)₂PbI₄, we explored two phase transitions occurring at different temperatures (Figure 4b,c). The steady-state reflectance maps showing these phase transitions for both compounds are provided in Figure S5, while more detailed VPVP measurement results at varying temperatures are presented in Figures S6–S8. Although the steady-state reflectance data (Figure S5) indicate that the phase transition-induced shifts in exciton resonance for these two compounds are not as pronounced as that observed for (BA)₂PbI₄, abrupt changes in reflectance are still discernible. From these, we were able to identify transient phase transitions in the VPVP experiments. Specifically, in the ΔR/R spectral maps (Figure 4a–c), the signature of phase transition is evident from the growth in ΔR/R amplitude during the first few microseconds.

In general, the VPVP data across different compounds suggest that phase transitions in these hybrid 2D MHPs are inherently slower than their 3D counterparts, such as CH₃NH₃PbI₃, which undergoes an orthorhombic-to-tetragonal phase transition within 100–200 ns at around 160 K. However, when comparing the phase transition kinetics of the measured 2D MHPs, as shown in Figure 4d, we observe that the phase transitions take place more rapidly at higher temperatures, including the transitions of (PA)₂PbI₄ at 310 K and (OA)₂PbI₄ at 300 K. This observation aligns with an Arrhenius-type dependence of phase transition on temperature and is consistent with previous findings of accelerated vibrational coupling between the organic and inorganic sublattices at elevated temperatures in MHPs. (41) Additionally, it mirrors the much faster (∼1 ns) tetragonal-to-cubic phase transition of CH₃NH₃PbI₃ at above room temperature, compared to its slower orthorhombic-to-tetragonal transition. (45)