Ultrafast transient infrared spectroscopy for probing trapping states in hybrid perovskite films

Hybrid perovskite materials exhibit exceptional optoelectronic properties but device performance and stability are limited by trap states that impede carrier transport. Conventional electrical techniques lack temporal resolution and carrier specificity, while common optical probes in the visible (time-resolved photoluminescence and transient absorption) do not provide direct spectral signatures of trapping, often yielding only multiexponential kinetics. The authors posit that femtosecond transient absorption in the mid-infrared (mid-IR), a spectral region dominated by free-electron absorption in the conduction band and devoid of overlapping organic vibrational modes, can directly monitor and distinguish electron recombination from trapping. The study aims to establish mid-IR TA as a sensitive, direct probe of trapping processes in hybrid perovskite thin films with different compositions, and to identify compositional, excitation, and probe-energy dependencies of trap formation and dynamics.

Prior femtosecond mid-IR studies have monitored free and trapped carriers in classical semiconductors (e.g., TiO2) where conduction-band electrons yield broad mid-IR absorption (∼3333–11,111 nm; 3000–900 cm−1), and IR emission features have been linked to electron trapping at shallow states. In hybrid perovskites, previous mid-IR work focused mainly on organic cation vibrational modes (e.g., NH stretches), not on free-electron signatures in a vibration-free window. Studies also indicate iodide-rich perovskites are more prone to defect formation and halide migration under illumination, potentially increasing trap densities. These reports motivate using a mid-IR window free of vibrational interference (4000–6000 nm; 2500–1666 cm−1) to directly observe trapping via changes in transient signal sign and dynamics.

Perovskite film preparation: CaF2 substrates were cleaned (DI water, acetone, IPA) and UV-ozone treated (10 min). FAPbBr3 and FAPbI3 films were prepared by a modified antisolvent dripping method: 1.1 M FAX and PbX3 (X = Br or I) in DMF:DMSO (9:1), spin-coated 15 s at 4000 rpm; 300 µL toluene dripped at 6 s; films annealed 10 min at 170 °C (FAPbI3) or 100 °C (FAPbBr3). Additional films included MAPbBr3, MAPbI3, and mixed-halide MAPbIxBr3−x with varying I/Br ratios. Substrates for reference included silicon wafers. Femtosecond transient absorption (fs-TA): Pump wavelengths from 410–530 nm were used; probe in the mid-IR from 4000–6000 nm (2500–1666 cm−1). Excitation powers typically 100–300 µW (varying with wavelength). Detection used an N2-cooled CCD sensitive to mid-IR photons. Visible-range fs-TA was also performed for comparison (e.g., monitoring ground-state bleach near 550–760 nm depending on composition). For MAPbI3, irradiation-time-dependent measurements were conducted by continuously illuminating the same spot and recording kinetics at different elapsed irradiation times; reversibility was tested by pausing irradiation. Kinetic traces were extracted at specific probe wavelengths (e.g., 4900, 5120, 4960, 4880 nm) and fitted with multiexponential functions to extract average lifetimes and appearance times for signal sign changes. Probe-wavelength dependence was assessed by holding pump at 520 nm and scanning probe energies (4000–6000 nm). Excitation-wavelength dependence was assessed by pumping at 410, 440, or 520 nm while probing at fixed mid-IR wavelengths (e.g., 4500 nm). A silicon wafer control was measured under identical conditions to exclude instrumental artifacts of negative signals.

• In the mid-IR window (4000–6000 nm), photoexcitation produces an initial positive transient signal attributed to conduction-band electron absorption. Emergence of a negative signal indicates mid-IR emission (gain) associated with emissive trapping states.
• MAPbBr3: Strong positive signal decays multi-exponentially with average lifetime ~120 ps at 4900 nm, followed by a small negative feature after >2 ns. Kinetics at 5120 nm show earlier sign inversion (~200 ps), demonstrating probe-wavelength sensitivity. Visible GSB kinetics (∼550 nm) match mid-IR positive decay but do not show sign inversion.
• MAPbI3: Mid-IR signal evolves with continuous irradiation. Initially (~0 min), positive signal persists to ~100 ps before a small negative appears; after 26 min irradiation, the positive converts to negative within ~10 ps. Turning off irradiation (~8 min) partially reverses the effect, indicating reversibility and a dynamical process linked to light-induced defect/trap formation. Visible TA near 760 nm shows no sign inversion; a silicon wafer control shows no negative mid-IR features.
• FA-based films: FAPbBr3 exhibits an initial near-time-zero negative feature converting to positive within ~60 fs (assigned to exciton thermalization/dissociation), then multi-exponential recombination with average lifetime ~10 ps and no persistent negative emission signature. FAPbI3 shows positive signal that converts to negative with a trapping time component ~8.5 ps; trapped electrons recombine slowly (>1 ns).
• Mixed-halide MAPbIxBr3−x: Increasing iodide content accelerates the appearance of the negative mid-IR signal at 4900 nm, with sign-inversion times ranging from ~1 ns (Br-rich) to ~100 ps (I-rich), demonstrating iodide control over emissive trap formation.
• Probe-wavelength dependence (MAPbI3, pump 520 nm): Sign inversion occurs later for higher-energy probe (4000 nm; switch ~500 ps) and earlier for lower-energy probe (6000 nm; switch ~20–22 ps; intermediate values: 5000 nm ~125 ps; 4500 nm ~43 ps). Higher-energy probes can liberate trapped electrons, delaying emission signatures.
• Excitation-wavelength dependence (MAPbI3, probe 4500 nm): Higher pump photon energy increases trapping probability and speeds sign inversion: 410 nm shows the fastest negative onset; 440 nm intermediate; 520 nm shows decay to zero with no negative signature despite higher power (∼300 µW), indicating energy, not intensity, governs trap population.
• Overall, iodide-containing and MA-based perovskites are more prone to form emissive trap states than Br/FA counterparts; mid-IR TA uniquely captures this via negative transient signals, unlike visible TA.

The study demonstrates that mid-IR fs-TA directly distinguishes electron recombination from trapping in hybrid perovskites by monitoring the sign of the transient signal: positive signals reflect conduction-band electron absorption, while negative signals reveal mid-IR gain associated with emissive trap states. Iodide content and the presence of MA cations promote trap formation, consistent with literature on halide-defect chemistry and ion migration. The probe-energy dependence shows that higher-energy mid-IR photons can free trapped electrons (reducing emission), whereas lower-energy probes facilitate stimulated IR emission from trapped carriers, explaining why visible probes often miss trapping signatures due to their much higher photon energies. Excitation-energy and light-soaking dependences link trap population and kinetics to photoinduced defect formation and ion migration. Thus, mid-IR TA provides a sensitive, selective window into trap-state dynamics that are otherwise convoluted with excitonic and free-carrier processes in the visible, directly addressing the need for an optical method to track and quantify trapping processes in perovskite films.

Transient absorption in the mid-IR uniquely tracks both free-electron dynamics and electron trapping in hybrid perovskite films. The emergence of a negative transient signal is assigned to mid-IR emission from emissive trap states, which are more prevalent in iodide- and MA-containing compositions. Trap signatures depend on film composition, probe wavelength (energy), excitation photon energy, and irradiation history, providing qualitative and quantitative handles to assess and control trap states. Visible-range TA lacks this specificity due to its propensity to liberate trapped carriers. These insights open avenues to engineer trap-state populations and explore mid-IR emission applications in iodide-based perovskites. Future work could identify the microscopic nature and energy distributions of emissive traps, map probe-energy thresholds for trap liberation, and correlate structural/processing variables with trap dynamics for device-relevant architectures.

• Potential hole contributions to mid-IR signals were assumed minimal but not fully ruled out; further studies are needed to confirm selectivity for electrons.
• Assignment of negative signals to emissive trap states is based on spectroscopy and comparative behavior; direct identification of trap species/energies is not provided.
• Measurements focus on a specific mid-IR window (4000–6000 nm); deeper spectral mapping could refine mechanisms.
• Film quality and light-soaking history affect trap formation; variability across preparation batches may influence generality.
• Visible TA does not show sign inversion, limiting cross-validation across spectral regions; complementary techniques (e.g., TRPL, THz) could strengthen assignments.