Tunable photostriction of halide perovskites through energy dependent photoexcitation

Halide perovskites exhibit giant photostriction, that is, volume or shape changes upon illumination. However, the microscopic origin of this phenomenon remains unclear and there are experimental reports of both light-induced lattice expansion and contraction. In this Letter, we establish a general method, based on first-principles calculations and molecular orbital theory, which provides a microscopic picture of photostriction in insulators based on the orbital characters of their electronic bands near the Fermi level. For lead-halide perovskites, we find that different valence states have different bonding characters, leading to opposing strengthening or weakening of bonds depending on the photoexcitation energy. The overall trend is that light induces lattice contraction at low excitation energies, while giant lattice expansion occurs at high excitation energies, rationalizing experimental reports.

Figure 1

(a) Crystal structure of $\mathrm{Cs}\mathrm{Pb}{\mathrm{Br}}_3$. Electronic structure of $\mathrm{Cs}\mathrm{Pb}{\mathrm{Br}}_3$ (b) without and (c) with spin-orbit coupling. The arrows indicate optical transition energies at different high-symmetry points.

Figure 2

(a) Selective excitation of the antibonding VB state and the nonbonding VB-1 state at the band gap of $\mathrm{Cs}\mathrm{Pb}{\mathrm{Br}}_3$. Evolution of the lattice constant of $\mathrm{Cs}\mathrm{Pb}{\mathrm{Br}}_3$, (b) with increasing photoexcited carrier density $n$ upon selective excitation and (c) with increasing electron/hole doping concentration $n$ that populates/depopulates the CB/VB states.

Figure 3

Photostriction of $\mathrm{Cs}\mathrm{Pb}{\mathrm{Br}}_3$ with and without spin-orbit coupling as a function of photoexcited carrier density.

Figure 4

Change of the lattice constant $a$ compared to the lattice constant in the dark $a_0$ of $\mathrm{Cs}\mathrm{Pb}{\mathrm{Cl}}_3$, $\mathrm{Cs}\mathrm{Pb}{\mathrm{Br}}_3$, and ${\mathrm{CsPbI}}_3$ as a function of excited carrier density. All the results are obtained without $G_0{W}_0$ corrections and without SOC.