Thermal history controls the optoelectronic response of lead halide perovskites through structure and dynamics

Scientific Impact Assessment

Core Contribution

This paper establishes that dynamic octahedral-tilt nanodomains — nanometre-scale, picosecond-lifetime regions of correlated tilts that locally break crystallographic symmetry — are a near-universal feature of lead halide perovskites (LHPs), and that their character is governed by three independent levers: A-site cation (symmetry/anisotropy: FA→MA→Cs increasing disorder), halide (dynamic-disorder magnitude: Cl→Br→I), and, most novelly, thermal history. The central conceptual advance is that cooling/heating rate alone can drive nominally identical compositions into distinct average crystallographic phases and distinct hidden local orders — with direct optoelectronic consequences, demonstrated in MAPbI₃ where heating rate changes PLQY across the phase transition. This reframes thermal history as a design/degradation variable on par with composition, a genuinely new angle in a mature field where stability discussions have centered on ionic migration within a fixed lattice.

Methodological Rigor

The methodological breadth is a major strength. The authors combine single-crystal X-ray diffuse scattering (multiple synchrotron beamlines at Diamond and PETRA III), single-crystal neutron diffuse scattering (ANSTO), machine-learning-assisted molecular dynamics with several purpose-built force fields (MACE, Allegro, NEP), a phenomenological octahedral-tilt model, DSC, and temperature-dependent hyperspectral PL. The cross-validation between experiment and MD is repeatedly demonstrated (folded Brillouin-zone comparisons, S(q,ω) energy-window decomposition). The resolution of the flower-like diffuse scattering into quasi-elastic (tilt) vs. inelastic 2–3 meV (Cs–Pb correlated displacement) channels is careful and addresses a long-standing off-centering debate directly. Alternative nanodomain models are explicitly tested and rejected. Weaknesses: the optoelectronic link (thermal history → PLQY → Voc) is the paper's headline claim but is demonstrated on a single composition (MAPbI₃) with modest ~1 K shifts and a single measured property; the leap to device Voc is inferential rather than demonstrated on actual devices. The connection of single-crystal cooling rates to IEC module thermal-cycling protocols is plausible but not directly validated in operating cells.

Potential Impact

The impact is potentially high. LHP photovoltaics/LED/detector research is a very large, active field, and this work supplies both a unifying structural framework and a practical new processing knob. If the thermal-history dependence of optoelectronic quality holds at the device level, it would influence fabrication protocols, stability-testing standards, and interpretation of reproducibility problems that have long plagued the field ("explains why perovskite performance depends so strongly on processing"). The compositional map of nanodomain character is a reference resource. The correction of prior powder-XRD interpretations (apparent phase coexistence being misread QEDS; CsPbBr₃ room-temperature Pnma assignment) has corrective value. Two open-source tools (rspace3d, gpuscatter) add reusable infrastructure.

Timeliness & Relevance

Highly timely. With PCE approaching the detailed-balance limit, stability is the acknowledged bottleneck, and intrinsic (non-chemical) structural instability is an underexplored axis. Interest in local/hidden order in functional materials (relaxors, batteries, cuprates) is surging, and this work connects LHPs to that broader diffuse-scattering community. Machine-learned potentials for phase-transition modeling are a current frontier, deployed here at scale.

Strengths & Limitations

Strengths: exceptional experimental scope (Cs/MA/FA × Cl/Br/I/mixed across all phases); rigorous experiment–simulation integration; resolution of two literature debates; a genuinely novel and actionable conceptual claim (thermal history as design variable); large international collaboration with strong facility access; open code and data commitments.

Limitations: the optoelectronic payoff — the part most likely to change practice — rests on a single-composition, single-property demonstration with small absolute effects; generalization to devices is asserted ("we expect this finding to be universal") rather than shown. Measurements are on single crystals, whereas devices use thin films with different defect landscapes and grain structure. The causal chain from nanodomain character to non-radiative recombination remains correlative. Some structural assignments (e.g., unconventional a⁰a⁰c⁺ MAPbI₃ at slow cooling) are unusual and would benefit from independent confirmation.

Other Observations

Reproducibility is aided by extensive Methods, named software, and committed data/code repositories, though the full reproduction requires substantial synchrotron/neutron beamtime and large-scale MD compute — a high barrier to entry. The work is strongly foundational in the sense of providing a framework and tools others will build on; multiple recent references are the authors' own prior work (Nature Nanotechnology 2025, arXiv preprints), indicating this is part of an active, self-reinforcing program. The interdisciplinary reach spans crystallography, computational materials science, photovoltaics/optoelectronics, and solid-state physics (diffuse scattering community).

Overall, this is a high-quality, broad, and conceptually fresh contribution that is likely to be widely cited and to shift how the field thinks about processing-dependent structure. Its ceiling on immediate practice-changing impact is set by the thin device-level demonstration of the optoelectronic consequences.