Tutorial: Transient Absorption Spectroscopy for Probing Ultrafast Dynamics

Abstract

Transient absorption spectroscopy (TAS) is an indispensable technique for directly tracking chemical and physical dynamics on ultrafast time scales. This Tutorial covers the basic principles behind TAS and illustrates with examples its application in probing and understanding ultrafast photophysical and photochemical processes in chemistry, physics, biochemistry, and materials science. We also discuss how TAS data are analyzed mathematically and kinetically to gain physical and mechanistic insights into complex dynamic processes. By providing practical guidance from experimental design to quantitative kinetic modeling, this Tutorial aims to equip researchers with the foundational knowledge to effectively utilize TAS for ultrafast studies of different processes in molecules and materials.

1. Introduction

1.1. Laser Technology Evolution from Continuous Wave Laser to Pulsed Laser

Understanding the efficiency of modern optoelectronic materials requires methods capable of capturing dynamics on their ultrafast time scales. Conventional spectroscopy, primarily utilizing broadband lamps, provides insights into steady-state properties such as electronic absorption and fluorescence spectra. For more specialized steady-state measurements requiring high spectral resolution or intensity, such as cavity output or Raman spectroscopy, continuous-wave (CW) lasers are frequently employed. , However, understanding dynamic processes, especially on ultrafast time scales, requires time-resolved techniques, such as ultrafast laser spectroscopy. Such measurements usually employ femtosecond or picosecond (ps) laser systems based on mode-locked oscillators generating pulses lasting tens to hundreds of femtoseconds (fs) in conjunction with regenerative femtosecond amplifiers. − The advancement in ultrafast laser spectroscopy has enabled the study of ultrafast photochemical and photophysical processes on the femtosecond, picosecond, or longer time scales and allows deeper fundamental understanding at an unprecedented level. −

1.2. The Core Challenge: Probing Ultrafast Carrier Dynamics

The primary event in any photophysical process is the nearly instantaneous excitation of the system to an excited electronic state by light, a vertical transition occurring according to the Franck–Condon principle. This is immediately followed by a cascade of processes including electronic and vibrational relaxation and recombination or charge or energy transfer, depending on the environment of the specific light-absorbing species or material lattice. Direct probing of the dynamic process is critical for both fundamental mechanistic understanding and applications. The earliest dynamics, spanning the attosecond (10^–18 s) to femtosecond time scale, encompass electronic dephasing and the thermalization of hot charge carriers as they dissipate excess energy into the crystal lattice. , Subsequent processes, occurring in the picosecond to nanosecond regime, are dictated by the competition between application-specific functional pathways (e.g., radiative recombination for luminescence or interfacial charge transfer for photovoltaics and photocatalysis) and competing loss mechanisms (e.g., nonradiative Auger recombination or carrier trapping at defect sites). A fundamental challenge in spectroscopy is that these competing pathways are often spectrally and temporally overlapping and convoluted. Therefore, part of ultrafast dynamics studies involve deconvolution of the multiple and complex processes through mathematical and kinetic modeling or computational studies.

1.3. Transient Absorption Spectroscopy

To address the requirement for high temporal resolution, transient absorption spectroscopy (TAS) has emerged as the definitive primary tool for this purpose. TAS is a powerful pump–probe technique that directly monitors photoinduced changes in a material’s optical absorption. This captures a real time observational window into the primary photophysical events and provides the spectral and kinetic information necessary for detailed material characterization, allowing for the precise identification of transient species and the accurate determination of their formation and decay rates, and enables the detailed elucidation of underlying photophysical mechanisms. −

1.4. Scope of This Tutorial

This Tutorial provides a guide to the principles and applications of TAS with several case studies as examples. We first introduce the fundamental concepts and experimental methods for TAS. Then, we discuss specific applications of TAS in studying ultrafast dynamics, detailing key experimental strategies and interpreting representative data from the recent literature. This Tutorial aims to provide fundamental knowledge to design TAS experiments and apply quantitative kinetic modeling to systems of interest.

2. Fundamental Principles Behind TAS

2.1. Pump–Probe Technique and the Concept of Optical Time Delay

The conceptual foundation of TAS originates from the “flash photolysis” technique developed by George Porter and Ronald Norrish in the 1950s, work that was recognized with the 1967 Nobel Prize in Chemistry. This early technique established the core principle of TAS, which instantaneously initiated a reaction and monitored the subsequent change in absorption. Modern TAS extends this principle into the ultrafast domain, utilizing synchronized femtosecond laser pulses to temporally probe excited-state dynamics with orders of magnitude faster time resolution than the original microsecond scale experiments.

As illustrated in Figure , the general layout of a pump–probe setup used in typical TAS experiments relies on two synchronized laser pulses. The first is the pump pulse, which excites the sample, and the second is the probe pulse, which interrogates the excited-state population by monitoring its change in absorbance. In this configuration, the pump beam is dumped to prevent pump scatter from reaching the detector, after passing through the sample, while the probe beam is directed to the detector for measurement.

1.

Illustration of the general idea of a pump–probe experiment.

Before detailing the instrument layout, it is essential to understand how time is measured in this ultrafast regime. Electronic detectors are generally too slow to resolve femtosecond events. Instead, TAS utilizes the constancy of the speed of light (c) to convert time to distance. The fundamental relationship between the time delay (Δt) and the difference in the physical path length (Δd) traveled by the pump and probe pulses is defined by

$$\mathrm{\Delta }t=\mathrm{\Delta }d/c$$ 1

Consequently, by using a motorized translation stage to increase the path length of the probe pulse in micrometer-scale increments, we can monitor the change in absorbance as a function of time with sub-picosecond precision.

2.2. TAS Experimental Setup and Key Components

To perform a TAS experiment, an ultrafast pulsed laser system is necessary. Figure shows an example of a specific femtosecond laser setup from light conversion. However, most TAS setups are similar in terms of their major components. The laser produces a beam using a diode-pumped solid-state design with Yb-doped gain medium. Diode lasers supply energy that excites the Yb ions and creates a population inversion inside a compact cavity. Through stimulated emission, this setup generates high-power femtosecond pulses (1030 nm), which can then be converted to other wavelengths by using harmonic modules. Among the most commonly used solid-state lasers are Ti:sapphire, Yb:KGW, and Nd:YAG. − For example, the specific system shown in Figure uses Yb:KGW to generate a 1030 nm fundamental beam with a 20 kHz repetition rate.

2.

Optical layout of light conversion TA setup. The experimental schematic illustrates the generation of tunable excitation pulses via an optical parametric amplifier (OPA) and the subsequent synchronization with a white-light continuum probe. A 9:1 prism bifurcates the fundamental laser output, directing the pump arm through a chopper (CH) for differential signal modulation and a delay stage (ΔT) to provide the temporal resolution. The probe arm undergoes supercontinuum generation in a nonlinear crystal (C) before both beams are focused onto the sample via a parabolic mirror (PM) and a lens (L), respectively. The transmitted probe is captured by the detector to monitor transient changes in optical density, while the residual pump is removed by a dump.

After the fundamental beam is generated, it is split into two distinct pump (blue) and probe (red) paths. Approximately 90% of the beam is directed into an Optical Parametric Amplifier (OPA). The OPA is a nonlinear optical device that mixes a broadband white light seed signal with the fundamental beam in a β-BaB₂O₄ (BBO) nonlinear crystal, generating two lower energy beams, called signal and idler, through a second-order (χ²) nonlinear interaction. By adjusting the tilting angle of the BBO crystal, one can tune which signal and idler wavelengths are phase-matched, thus generating a tunable pump wavelength. As the tuned single-wavelength pump beam exits the OPA, it passes through several essential optics. The chopper (CH) modulates the pump beam by blocking every other pulse and acts as an on/off modulator, allowing the instrument to record the difference in probe absorption. This is followed by a half-waveplate (λ/2), which rotates the polarization of the pump beam to the magic angle (54.7°) relative to the probe beam. This configuration eliminates kinetic artifacts arising from molecular rotational diffusion following the initial photoselection, ensuring that only pure population dynamics are measured. A neutral density filter (NDF) is used to attenuate the pump power, ensuring the experiment remains in the linear excitation regime, while a color filter (CF) removes any residual fundamental or unwanted light from the OPA output.

The remaining 10% of the fundamental beam is directed to a motorized delay stage and then into a nonlinear sapphire crystal (C) to undergo supercontinuum generation, transforming the single-wavelength fundamental into a broadband white light continuum. A filter (F) is used to remove the remaining 1030 nm light, leaving only the white light probe. A parabolic mirror (PM) is used to focus the probe onto the sample; unlike a standard lens, a PM mirror focuses all wavelengths of white light to the same point without chromatic aberration. The pump and probe beams were spatially and temporally overlapped on the sample. After interaction with the sample, the pump beam is directed to a beam dump, while the probe beam is directed to the detector to record the transmission spectra.

2.3. Transient Absorption Signals

The fundamental observation is the difference in absorbance (ΔA), which represents the change in the sample’s optical density induced by the pump pulse. Mathematically, this is defined by the logarithmic ratio of the transmitted probe intensities in the absence and presence of the pump:

$$\mathrm{\Delta }A(\lambda ,t)=\log \left(\frac{I_{\mathrm{u}\mathrm{n}\mathrm{p}\mathrm{u}\mathrm{m}\mathrm{p}\mathrm{e}\mathrm{d}}(\lambda )}{I_{\mathrm{p}\mathrm{u}\mathrm{m}\mathrm{p}\mathrm{e}\mathrm{d}}(\lambda ,t)}\right)$$ 2

By measuring this difference and systematically varying the time delay (t) between the pump and probe pulses, one constructs a two-dimensional map ΔA(λ,t). This map captures the temporal evolution of photoinduced processes such as energy relaxation, charge transfer, intersystem crossing, and the formation of nonemissive (dark) states. The observed (ΔA) spectrum is a linear superposition of three distinct physical contributions. Understanding these three elements is essential for interpreting the data, as they correspond to specific changes in the population and electronic state of the sample, as illustrated in Figure .

3.

An illustration of pump–probe dynamics of a model system. The pump pulse promotes the system to the excited state from which the excited species may decay to the valence band. a) A simplified potential energy surface diagram of the photoreaction. b) Illustration of the isolated constituents of the TA/TB spectrum and their gross product.

Figure a illustrates the photoreaction through a simplified potential energy surface (PES) diagram. The process begins with a pump pulse (solid black arrow) that promotes the system from the ground state (GS) to an excited state (ES). This excitation immediately depletes the GS population, which is the physical origin of the transient bleaching (TB) signal. This bleach persists as long as the ground state remains depopulated. Following excitation, the probe pulse monitors the system’s evolution by interrogating the new electronic populations. The probe may be absorbed by species in the ES, promoting them to a higher excited state (HES, blue arrow) and resulting in transient absorption (TA). Additionally, the probe field can induce a transition from the ES back to the GS (green arrow), manifesting as stimulated emission (SE). It is vital to distinguish SE from general relaxation pathways, such as nonradiative decay; while various processes contribute to the repopulation of the ground state (driving TB recovery), only the specific probe-induced transition generates the SE signal in the measured data. Figure b displays the resulting signals generated in a TAS experiment. The solid line represents the total signal visualized by the spectrometer, which is the sum of the overlapping TB, SE, and TA components. Because these signals often overlap in both wavelength and time, they can partially cancel each other out, complicating the direct extraction of lifetimes from raw data.

The TB, also known as the ground state bleach (GSB), is one major component of a TAS spectrum. Physically, the pump pulse depletes the population of the GS, resulting in fewer molecules or electrons available to absorb the probe light compared to the unpumped sample. Consequently, the sample appears more transparent (bleached) at wavelengths where the GS species absorb, leading to an increase in transmitted probe intensity (I _pumped > I _unpumped). The recovery of the TB signal over time corresponds to the repopulation of the GS, as represented by the orange-shaded regions in Figure .

TA is another major component. A TA signal, often termed photoinduced absorption (PA) or excited-state absorption (ESA), is observed when the population in the ES absorbs a probe photon to transition to HES. In this case, the sample absorbs more probe light than it did in the GS, resulting in a decrease in transmitted probe intensity (I _pumped < I _unpumped). The TA signal is critical for identifying nonemissive states, reaction intermediates, and charge-separated species that are otherwise invisible to steady-state techniques, represented by the blue-shaded regions in Figure .

Another possible contribution to the transient signal is the SE. This may occur when the probe wavelength overlaps with the sample’s fluorescence emission spectrum and when the excited-state lifetime is relatively long and fluorescence quantum yield is high. If the probe photon energy matches the energy difference between the ES and the GS, then the probe field could induce a transition back to the GS via SE. The transition probabilities for absorption (from the GS to the ES) and SE (from the ES to the GS) are equal as described by Einstein coefficients, − implying that, once the excited state is populated, the probe pulse could induce stimulated emission back to the GS. This process results in the emission of a photon coherent with the probe beam, which adds to the probe intensity, reaching the detector. Like TB, SE results in an increase in transmitted probe intensity (I _pumped > I _unpumped). SE is typically Stokes-shifted (red-shifted) relative to the TB band and tracks the population of the emissive excited state, as represented by the green-shaded regions in Figure .

2.4. Pulse Duration and Time Resolution

The temporal resolution of a TAS experiment is fundamentally limited by the duration of the laser pulses used. To date, the shortest laser pulses are reported as 19.2 attoseconds (as). However, in more common femtosecond TAS applications, pulse durations typically range from 3.3 fs up to 100 fs. This selection involves an inherent physical trade-off known as the time-bandwidth product. For a transform-limited Gaussian pulse, the relationship between the temporal width (Δt) and the spectral frequency width (Δv) follows the inequality ΔtΔv ≥ 0.441. , Consequently, achieving a high temporal resolution (small Δt) necessitates a broad spectral bandwidth (large Δv). For example, while 10 fs pulses provide excellent time resolution, their broad bandwidth (∼100 nm) may simultaneously excite multiple overlapping electronic states. If spectral selectivity is required to distinguish between specific species, a longer pulse (∼30–100 fs) may be preferable.

Beyond the intrinsic pulse width, the effective time resolution is significantly impacted by group velocity dispersion (GVD). This phenomenon occurs as the broadband probe pulse traverses dispersive optical media, including the sample cuvette, refractive lenses, and nonlinear crystals required for supercontinuum generation. This results in a temporally “chirped” probe, where redder wavelengths arrive at the sample earlier than bluer ones. To mitigate this effect, dispersive-free optical designs, such as PM, are commonly used, although residual chirp must be computationally corrected during the data processing stage to resolve the true time-zero across all wavelengths.

The overall temporal performance of the TAS system is characterized by the instrument response function (IRF), which is typically modeled as a Gaussian function with a full-width at half-maximum (fwhm) determined by the cross-correlation of the pump and probe pulses. In a well-optimized setup, the IRF may reach values as low as 50–120 fs. ,,

Crucially, during the initial period of pulse overlap (at or near t = 0), the TA signal is often contaminated by nonresonant coherent artifacts, such as cross-phase modulation and two-photon absorption from the solvent or substrate. To obtain meaningful kinetic information, these artifacts must be effectively disentangled or subtracted from the transient spectra.

2.5. Pulse Fluence and Excitation Density

The intensity of the pump pulse, or fluence (measured in μj/cm²), is a critical parameter that dictates the regime of the measured dynamics for the initial population density of excited states. This parameter determines whether the experiment probes intrinsic single-particle dynamics or collective multiparticle interactions. In the linear regime, excitations are spatially sparse and noninteracting. Consequently, the transient signal amplitude scales linearly with pump power, and the decay kinetics remain independent of excitation density. Operating in this limit is strictly required to accurately determine intrinsic material parameters such as carrier lifetimes, trapping rates, and electron–phonon coupling constants.

On the other hand, the nonlinear regime emerges at higher pump fluences where excitation densities are sufficient to drive particle–particle interactions. Rather than being strictly avoided as artifacts, these nonlinearities offer a direct probe into complex scattering and recombination mechanisms. In semiconductors, high carrier densities lead to Auger recombination, where the energy of an electron–hole recombination is nonradiatively transferred to a third carrier. Similarly, in molecular systems, exciton–exciton annihilation can occur when two adjacent excited states interact, resulting in one returning to the ground state. These many-body processes manifest as fast, fluence-dependent decay components that do not represent the intrinsic low-density dynamics. Therefore, performing a systematic power dependence study enables the researcher to map the kinetic order of the decay, separating intrinsic monomolecular events from high-order recombination pathways, though care must always be taken to stay below the threshold of sample photodegradation.

3. Data Analysis of TAS

Quantitative analysis of the multidimensional transient absorption data set, ΔA(λ, t), is essential for translating spectroscopic observation into mechanistic understanding. This process requires advanced methods to separate the spectral contributions and kinetics of the overlapping transient species. The analysis proceeds hierarchically, beginning with dimensionality reduction and qualitative inspection, followed by global analysis to establish time scales, and concluding with target analysis to resolve the physical reaction mechanism. ,

3.1. Data Validation and Chirp Correction

Before kinetic parameters can be extracted, the raw data must be validated and corrected for physical artifacts. As discussed in Section , the most critical correction is for group velocity dispersion (GVD) or “chirp”. Because the blue components of the broadband probe travel slower than the red components, time zero (t = 0) is not constant across the spectrum, as shown in Figure a. A computational correction is applied using the coherent artifact signal recorded in the blank to realign the signal onset, ensuring that t = 0 is simultaneous for all wavelengths, as shown in Figure b.

4.

Comparison of raw and computational corrected 2D TA map. (a) Raw 2D map showing the characteristic spectral chirp (dashed line) caused by GVD. (b) The same data set after computational correction, ensuring that t = 0 is simultaneous for all wavelengths.

3.2. Dimensionality Reduction via Singular Value Decomposition

A single TAS experiment generates a massive data matrix, containing thousands of kinetic traces. While it is common practice to focus on a single diagnostic wavelength for analysis, individual traces are often highly susceptible to random experimental noise such as laser instability or detector electronic fluctuations. This noise can mask the kinetic features, making single-wavelength analysis unreliable. To manage this, singular value decomposition (SVD) is employed as a diagnostic and noise suppression tool. , SVD decomposes the complex data matrix into orthogonal components, separating dominant spectral features from random experimental noise. In addition, SVD helps to estimate the rank of the system, which corresponds to the number of distinct kinetic components required to describe the data. Knowing the approximate rank provides a robust foundation for the selection of an appropriate kinetic model.

3.3. Single-Wavelength Kinetic Analysis

Quantitative analysis typically begins with fitting the kinetic traces to extract rate constants. For single-wavelength kinetics, the time-dependent signal I(t) is typically modeled using a single or multiexponential decay function convoluted with the IRF:

$$I(t)=\mathrm{I}\mathrm{R}\mathrm{F}(t)\otimes \sum _{i=1}^N{A}_i{\mathrm{e}}^{-t/{\tau }_i}$$ 3

Here, τ _i represents the lifetime of the i-th component and A _i is its amplitude. While useful for simple systems, single wavelength fitting often yields ambiguous results when spectral features overlap, as the extracted lifetimes may represent a composite of multiple coinciding processes rather than distinct species. ,,

3.4. Global Analysis: Establishing Systemic Time Scales

To resolve complex dynamics, global analysis is applied to the entire data set simultaneously. This method decomposes the data matrix assuming that the kinetics at all wavelengths can be described by a shared set of global rate constants (k _i ) or lifetimes (τ _i = 1/k _i ). , Because the data set is treated as a matrix of time-dependent spectra, the type of output depends on the model applied. When a parallel kinetic model (sum of independent decays) is used, the analysis yields decay-associated difference spectra (DADS). DADS represent the spectral amplitude associated with each decay lifetime. Alternatively, when a sequential kinetic model is applied (e.g., State 1 → State 2 → State 3), the result is a set of evolution-associated difference spectra (EADS). ,, EADS visualize how the total spectrum evolves through consecutive kinetic windows defined by these global lifetimes. However, this sequential model is often a mathematical abstraction, which assumes a unidirectional flow of population and cannot distinguish between competing pathways or equilibria, resulting in spectra that may be linear combinations of multiple species.

3.5. Target Analysis: Resolving the Physical Mechanism

To determine the specific reaction mechanism, target analysis is applied by imposing physical constraints upon the kinetic fitting. In this approach, the researcher postulates a kinetic model based on prior chemical knowledge to describe how the system evolves. The process involves defining a kinetic scheme that specifies the postulated photochemical or photophysical pathways, including complex routes like forward, reverse, and branching reactions between the proposed chemical intermediates. , This scheme is expressed mathematically by a connectivity matrix (K-matrix), which governs the population dynamics (c _i ) of each transient species via a system of coupled differential equations:

$$\frac{\mathrm{d}{c}_i}{\mathrm{d}t}=\sum _{j=1}^n{k}_{j\to i}{c}_j(t)-\sum _{l=1}^n{k}_{i\to l}{c}_i(t)$$ 4

In this equation, k _{j →i} and k _{i →l} represent the microscopic rate constants for the population transfer between species, which are physically distinct from the observed global lifetimes (τ _i ). By fitting the data to this specific mechanism, the analysis yields species-associated difference spectra (SADS), , which represent the estimated pure spectra of the intermediates, along with the microscopic rate constants for each step. It is important to note that both target and global methods generally assume bilinearity, meaning the spectral shape of each species remains constant over time.

3.6. Physical Interpretation: From Rate Constants to Mechanisms

The mathematical rigor of global and target analyses yields a set of microscopic rate constants (k) and SADS. The final step is to translate these mathematical parameters into concrete physical mechanisms. This involves mapping the extracted time constants onto the material’s electronic band structure (for semiconductors) or potential energy surface (for molecules). By correlating the magnitude of the rate constants with the spectral evolution observed in the EADS and SADS, the dynamics are typically categorized into three temporal regimes.

The ultrafast regime (<1 ps) captures the primary photophysical events immediately following pump excitation. This period is dominated by electronic dephasing, where the initial phase relationship established between electronic states is lost due to environmental interactions. In semiconductors, pump photons often impart energy in excess of the bandgap to generate hot carriers. These carriers rapidly dissipate excess energy into the crystal lattice through carrier–carrier scattering, involving collisions among the excited charges, and electron–phonon coupling, eventually thermalizing to the band edge. − In molecular systems, this regime corresponds to internal conversion (IC) and relaxation from the initially Franck–Condon region. This represents the nonequilibrium nuclear configuration of the ES immediately following a vertical transition, where the system retains the ground-state geometry before nuclear rearrangement begins. IC involves the relaxation of higher excited singlet states to the lowest excited state (S _n → S ₁). Furthermore, in transition metal complexes such as iron or ruthenium polypyridyls, strong spin–orbit coupling drives Intersystem Crossing (ISC) from singlet to triplet states within tens of femtoseconds. This behavior defies the classical view of ISC as a slow process and competes directly with vibrational relaxation. −

The intermediate regime (picoseconds to nanoseconds) is defined by the reorganization of the system’s nuclear framework and its environment, encompassing vibrational cooling, solvation, and trapping. − As the system evolves beyond the initial ultrafast scattering events, it enters a phase of reorganization where the excited species interacts with its surroundings to dissipate excess energy and stabilize its new electronic configuration. Rather than a single dominant mechanism, this regime is often characterized by competition among internal relaxation, environmental response, and carrier transport. In molecular systems, this typically involves the rotational and translational reorientation of the solvent shell to accommodate changes in the solute dipole moment, often manifested as dynamic spectral shifts or band narrowing in the transient absorption signal. In nanostructured materials, such as perovskites, this regime is often dominated by charge carrier diffusion and migration. During this period, excited carriers may drift toward interfaces for extraction or be captured by trap states originating from crystal defects. ,

The slow regime (ns to μs) corresponds to the decay of metastable states or the formation of reaction products, dominated by recombination. Signals in this regime often arise from TA of long-lived species such as charge-separated states, triplet excitons in organic semiconductors, or chemical photoproducts. Ultimately, the system returns to the ground state via radiative recombination, such as fluorescence or phosphorescence, or via nonradiative recombination. , The longevity of the signal in this regime defines the carrier lifetime, representing the window of opportunity for the material to perform useful applications. For instance, in solar cells, carriers must be extracted to the electrodes before they recombine, while in photocatalysis, the excited state must persist long enough to drive a redox reaction at the interface. ,,

4. Case Studies of TAS

In this section, we present three case studies that illustrate the transition from raw dynamic data to physical mechanisms. These examples demonstrate the relative strengths of specific mathematical fitting, global analysis, and target modeling across diverse systems.

4.1. Wavelength-Independent Hot Electron Relaxation in Plasmonic Gold Nanostructures

The investigation of nonequilibrium carrier dynamics in plasmonic nanostructures provides a foundational perspective on how TAS is used to isolate fundamental physical rates from experimental variables. − Noble metal nanoparticles, such as gold nanospheres (AuNSs) and gold nanorods (AuNRs), , serve as quintessential model systems because their localized surface plasmon resonance (LSPR) recovery directly tracks the dissipation of energy from hot electrons to the lattice via electron–phonon coupling. , When these particles interact with light, the collective oscillation of conduction electrons results in a nonequilibrium hot electron distribution. , These hot carriers are intensely discussed for applications in energy conversion and catalysis, where their reactivity is highly dependent on how they are generated and how they thermalize. Because the decay of a plasmon into these hot electrons is sensitive to the photon energy, ultrafast studies are required to distinguish between different excitation mechanisms, such as interband (from the 5d band to the 6sp band) and intraband transitions. , Investigating these dynamics across different particle sizes and wavelengths allows for the rational design of plasmonic systems with an optimized hot-carrier generation efficiency.

Figure a illustrates the 3D TA map of 30 nm AuNPs showing the spectral evolution over time. The differential absorption spectra, as shown in Figure b, reveal a characteristic negative bleach signal centered at approximately 525 nm, which corresponds to the broadened and shifted LSPR peak following photoexcitation. This is accompanied by two positive sidebands resulting from the change in dielectric constants that broaden the plasmon absorption. ,, As the bleach minimum shifts spectrally during cooling, monitoring intensity at a single wavelength introduces artifacts; accurate kinetic data therefore require extraction of the peak amplitude via full spectral fitting at each time delay.

5.

(a) 2D TA map of 30 nm AuNPs excited with 100 μJ/cm2 pulses at 400 nm. (b) TA spectra at different times after excitation, where t _max represents the time at maximum bleach intensity. Reproduced from ref . Copyright 2017 American Chemical Society.

The temporal evolution of this bleach intensity is typically analyzed using mathematical fitting to extract specific lifetimes. A facile approach is the use of a biexponential decay, where a fast component represents the electron–phonon (e–ph) coupling and a slow component describes the coupling of the nanoparticle to the surrounding environment. While the relaxation rates in the 10–60 nm range appear size-independent, ,, power-dependent analysis reveals a linear relationship between the e–ph coupling time and pump intensity. This linearity arises because the electron heat capacity increases with temperature while the coupling strength remains constant, rendering the e–ph coupling time a direct metric for the initial electron temperature. ,,

To further explore the efficiency of hot-electron generation, researchers can employ more specialized kinetic analysis. By extrapolating e–ph coupling times to zero pump intensity, it is possible to determine the internal e–ph coupling time (τ_el–ph ≈ 0.9 ps), which remains consistent regardless of whether interband transitions are involved. This consistency suggests that, while the initial distribution may differ, the fundamental decay process into the lattice is unchanged. The slope of the e–ph coupling time versus pump intensity then serves as a metric for how effectively a specific wavelength generates hot carriers.

As shown in Figure a,b, the hot-electron generation efficiency clusters into two distinct regimes based on the interband transition threshold. When the excitation wavelength is below 550 nm (E > 2.25 eV), direct interband transitions from the 5d to the 6sp band are possible. In this regime, the slopes are approximately twice as high as those in the intraband regime (>550 nm), where carriers are generated solely via plasmon decay. This indicates that interband excitation is significantly more effective at generating hot electrons for a given pump intensity. Furthermore, the absence of size dependence in these regimes supports the theory that the plasmon provides the necessary momentum for intraband transitions, whereas surface scattering is less relevant in the 12–56 nm range.

6.

(a) e–ph coupling times of 30 m AuNPs excited with different, fixed wavelengths. The solid lines are linear fits. (b) Slopes of the dependences observed in panel a. The red boxes highlight the two relevant regimes. Reproduced from ref . Copyright 2017 American Chemical Society.

Overall, this case study establishes that the choice of excitation wavelength relative to the interband threshold is a critical factor for maximizing hot-carrier yields in metallic nanoparticles. While simple mathematical fitting, such as biexponential decay, identifies the intrinsic cooling time, analyzing the intensity-dependent slopes reveals the superior efficiency of interband transitions (wavelengths <550 nm) for generating hot carriers. These insights are essential for the rational design of plasmonic systems in photocatalysis and photovoltaic applications.

4.2. Ultrafast Charge Carrier Dynamics of CH3NH₃PbBr₃ Perovskite Nanoclusters

Perovskites are a class of materials that have generated high interest in recent years due to their attractive optoelectronic properties. − Recently, highly confined perovskite magic-sized clusters (PMSCs) have been identified as stable intermediates of interest in understanding the growth mechanisms of perovskite quantum dots (PQDs). − In addition, due to their small size relative to PQDs and to perovskite bulk, PMSCs have been proposed as a useful model system for studying the effect of defects on the optoelectronic properties of electrons due to their large surface-to-volume (S/V) ratio. − Because of this large S/V ratio compared with those of PQDs and perovskite bulk, PMSCs possess a higher degree of surface defects, which can contribute to electron/hole trapping and nonradiative exciton recombination. − Ultrafast studies are therefore essential to investigate the specific carrier dynamics of these novel nanoclusters and to assess the impact of incomplete surface passivation on their performance.

Figure a illustrates the TA spectra of the PMSCs at various time delays. The spectra exhibit three distinct features: two photoinduced absorption (PA) peaks centered at 420 nm (PA₁) and 461 nm (PA₂) and a photoinduced bleach (PB) feature peaking at 448 nm. Mathematical fitting, as shown in Figure b, determines the time constants and amplitudes of each decay component.

7.

(a) Difference in absorption spectrum as a function of wavelength and varying probe delay times and (b) overlaid kinetic decay profiles and fits of the PA₁, PB, and PA₂ signals. Reproduced from ref . Copyright 2024 American Chemical Society.

Fitting the decay profiles in this way allows for assumptions to be made about the path of the electron through various states based on the number and time scale of the decay components. A weakness of this fitting strategy is that the extracted exponential components consider only the overall forward process and do not give information about possible reverse processes. In addition, while a general claim can be made regarding the involved states, the absence of more complex modeling requires that the involvement of any specific proposed states relies only on assumption. For example, in a decay profile like those shown in Figure b, which are fit to a triple exponential function, one may identify the fast component as cooling of the photoexcited electron to the conduction band edge or to trapping, the medium component to trap-assisted recombination, and the long component to band-to-band recombination, with the support of similar work. Overall, this mathematical fitting process is a facile method of TAS analysis that can give some information regarding the exciton dynamics of a system at the cost of the accuracy of the proposed model.

The limited insights gained by mathematical single-wavelength fitting can be enhanced using kinetic target analysis, which can be used to propose a detailed model that specifically defines the states, along with forward and reverse rates, which may be involved in the exciton recombination process. − The kinetic modeling process works by first establishing a proposed set of states, including the valence and conduction bands (VB and CB), band-edge states, and trap states. Then, a set of rate equations is defined to express the population of each proposed state as a function of the time, and rate constants for the transitions between each defined state are selected and iterated until a good fit to the experimental decay curve is achieved. An ideal model will achieve a good fit to the experimental decay profile with the fewest possible number of states and processes, similar to the simple mathematical fitting process that seeks to achieve a good fit with the fewest possible number of exponential components.

Figure a shows a possible model of the key processes involved in exciton recombination in PMSCs. Figure b shows the population of each identified state vs time, and Figure c shows the experimental decay profile fit to the kinetic modeling.

8.

(a) Diagram of proposed kinetic model and energy levels, (b) plot of population vs time, and (c) experimental data fit to kinetic fit trace. Reproduced from ref . Copyright 2024 American Chemical Society.

Here, as summarized in Figure a, several states have been proposed by kinetic modeling and include the CB, the CB edge (CB_edge), shallow trap (ST), deep trap (DT), and the VB. Between these states, eight processes have been identified: six forward processes and two reverse processes. From these proposed states and rates, the population of each state can be plotted over time, as shown in Figure b. In this model, the electron is defined to originate in the CB (i.e., the modeling process begins after photoexcitation of the electron from VB to CB). As the population of the CB declines, the populations of the CB_edge, ST, and DT states initially increase and then decrease over time, while the population of the VB steadily increases over time. This is consistent with the physical process of electron relaxation, where an initially excited electron in the CB may pass through various competing relaxation routes before ultimately being terminated in the VB. In addition to visualizing the population over time, the veracity of the proposed kinetic model is tested by fitting the calculated decay profile to the experimental data. This fit is shown in Figure c and shows a close fit to the experimental data, indicating that the proposed model is reasonable. Overall, this results in a more accurate and informative model of the carrier dynamics compared to the simple mathematical fitting process.

Once the kinetic fitting process is complete, the results of the modeling process can be utilized in conjunction with supplementary data to generate a final proposed model that includes involved states and forward and reverse processes and identifies the time scale of each process. The final proposed model in this case study is listed in Figure .

9.

Final proposed model based on kinetic and mathematical fitting. Reproduced from ref . Copyright 2024 American Chemical Society.

As summarized in Figure , photoexcited electrons vibrationally relax from the CB to the CB_edge in about 8 ps, which involves a forward process of 2 ps and a reverse process of 6.6 ps. Subsequently, the electron can undergo one of the following processes. It may further relax to the ST state over the course of 60 ps with kinetic modeling identifying a forward process of 10 ps and a competing reverse process of 50 ps with subsequent relaxation to the DT in 40 ps. Alternatively, the electron can relax directly from the CB_edge to the DT state in 666 ps, followed by recombination with the hole in the VB over an additional 6.6 ns. The last identified process is electron–hole recombination from CB_edge to VB through direct band edge recombination, a process that occurs over the course of about 93 ns.

4.3. Global Analysis of Photochemistry Dynamics of Anion Channelrhodopsin PsACR1

While target analysis is essential for branched, competitive kinetics, many biological systems function through defined sequential cycles. The anion channelrhodopsin PsACR1 is a prime example, serving as a critical tool for optogenetics due to its red-shifted absorption and fast channel-closing kinetics. , The functional cycle of this protein involves a cascade of activation and deactivation events spanning from femtoseconds to seconds, providing real-time observation to resolve the photochemical mechanism. −

A major analytical challenge in this system is the severe spectral congestion. As shown in Figure , the transient spectra are dominated by overlapping features, including an ESA at 480 nm, a GSB near 560 nm, and an SE at 720 nm. As the photocycle proceeds, positive absorption bands from the J (610 nm) and K (600 nm) intermediates emerge and superimpose directly onto the decaying ESA and GSB signals, making single-wavelength analysis intractable.

10.

Femtosecond-to-picosecond TAS of wild-type PsACR1 at pH 8. Selected TA spectra up to 29 ps. The photointermediates “J” and “K” are indicated with positive bands at 610 and 600 nm, respectively. ESA, GSB, and SE stand for excited-state absorption, ground-state bleach, and stimulated emission, respectively. The spectral region of 540–560 nm is omitted due to the strong pump-light scattering. Reproduced with permission from ref . Copyright 2017 Royal Society of Chemistry.

To deconvolute these overlapping contributions, global analysis was employed using a sequential interconverting kinetic model (State 1 → State 2 → State 3→ ···). In this framework, the system is modeled as a unidirectional cascade where each time constant corresponds to the lifetime of a specific EADS, shown in Figure a. Ideally, dimensionality reduction via SVD is used to determine the number of kinetic components objectively. For PsACR1, a comprehensive global fit required nine exponential components with time constants ranging from 100 fs to infinity to describe the data fully. The necessity of this approach is clearly illustrated by the time traces in Figure b. These traces, taken at 462 nm (blue), 560 nm (green), 630 nm (red), and 720 nm (gray), represent the raw, unfiltered evolution of the ESA, GSB, and PA. These spectra isolate the unique signature of each intermediate state free from the interference of concurrent transitions, providing clean snapshots of the protein’s evolution.

11.

Kinetic analysis and deconvolution of PsACR1 photodynamics. (a) EADS extracted from the fs–ps TA data. The gray curve represents the inverted normalized dark-state absorption for comparison with the GSB. The 540–560 nm region is omitted due to pump scattering. (b) Representative time traces up to 10 ps recorded at 462 nm (blue, ESA), 560 nm (green, GSB), 630 nm (red, photoproduct absorption), and 720 nm (gray, SE). Time constants (τ) derived from the global fit are indicated by gray arrows. Reproduced with permission from ref . Copyright 2017 Royal Society of Chemistry.

The physical mechanism of the photocycle is elucidated by tracing the evolution of the extracted EADS. , The first EADS, representing the time zero state, characterizes the protein immediately following excitation in the ES, exhibiting distinct ESA and SE features. The rapid transition from EADS 1 to EADS 2 within 100 fs reveals a blue-shift of the ESA to ∼460 nm and a red-shift of the SE, attributed to initial relaxation from the initial nonequilibrium ES at the Franck–Condon geometry. Subsequently, the formation of the J intermediate occurs within 500 fs, marked by the rise of the positive absorption band at 610 nm. This step corresponds to the primary photoisomerization of the retinal chromophore from an all-trans to a 13-cis configuration. The J intermediate then evolves into the K intermediate over 2.2 ps, a process physically assigned to vibrational relaxation (cooling) and evidenced by a 10 nm blue-shift of the product absorption. On much longer time scales, the analysis resolves the deprotonation of the retinal Schiff base (RSB) in 26 ms to form the M-intermediate. This specific step is correlated with channel closing and is followed by reprotonation in 83 ms. , The complete molecular map linking these calculated time constants to specific chemical states (J, K, L, M) is summarized in Figure of the original study, providing a robust mechanistic model of the biological function.

12.

Complete photocycle model of wild-type PsACR1 at pH 8 from femtoseconds to seconds. Proposed hydrogen-bonding interaction near the retinal Schiff base in the K intermediate and a proton transfer reaction are shown (balloons). Reproduced with permission from ref . Copyright 2017 Royal Society of Chemistry.

By using global analysis to deconvolute the data, we move from a complex set of overlapping curves to a robust molecular map. This allows us to link specific physical events (such as a bond flipping in picoseconds) to critical biological functions (such as a channel closing in milliseconds), providing the insights needed for rational optogenetic design.

5. Summary and Outlook

This Tutorial provides an overview of TAS as a powerful technique to investigate chemical and physical dynamics on ultrafast time scales, discusses the fundamental principles of the pump–probe technique, and details the experimental configurations required to achieve femtosecond temporal resolution. In addition, important mathematical and kinetic data analysis techniques were discussed, demonstrating how global and target analysis can separate overlapping signals to reveal the intrinsic time scales and pathways of a process. Through case studies of gold nanoparticles, perovskite nanoclusters, and biological proteins, we have shown how these analytical strategies transform complex transient signals into a complete mechanistic landscape. Overall, this Tutorial provides the foundation needed to design TAS experiments, interpret data, and apply insights to diverse cases in disciplines ranging from biochemistry to materials science.

Glossary

Abbreviations

AuNP

Gold Nanoparticle

AuNR

Gold Nanorod

AuNS

Gold Nanosphere

BBO

β-BaB₂O₄

C

Nonlinear Crystal

CB

Conduction Band

CB edge

CB_edge

CF

Color Filter

CH

Chopper

CW

Continuous Wave

DADS

Decay-Associated Difference Spectra

DT

Deep Trap

EADS

Evolution-Associated Difference Spectra

e–ph

electron–phonon

ES

Excited State

ESA

Excited-State Absorption

fs

Femtosecond

fwhm

Full-Width at Half-Maximum

GS

Ground State

GSB

Ground State Bleach

GVD

Group Velocity Dispersion

HES

Higher Excited State

IC

Internal Conversion

IRF

Instrument Response Function

ISC

Intersystem Crossing

L

Lens

LSPR

Localized Surface Plasmon Resonance

Nd:YAG

Neodymium-doped Yttrium Aluminum Garnet

NDF

Neutral Density Filter

ns

Nanosecond

OPA

Optical Parametric Amplifier

PA

Photoinduced Absorption

PB

Photoinduced Bleach

PES

Potential Energy Surface

PM

Parabolic Mirror

PMSCs

Perovskite Magic Sized Clusters

PQD

Perovskite Quantum Dot

ps

Picosecond

PsACR1

Proteomonas sulcata Anion Channelrhodopsin 1

RSB

Retinal Schiff Base

S/V

Surface to Volume

SADS

Species-Associated Difference Spectra

SE

Stimulated Emission

ST

Shallow Trap

SVD

Singular Value Decomposition

TA

Transient Absorption

TAS

Transient Absorption Spectroscopy

TB

Transient Bleaching

Ti:sapphire

Titanium-doped Sapphire

VB

Valence Band

Yb:KGW

Ytterbium-doped Potassium Gadolinium Tungstate

The manuscript was written through the collective contributions of all authors. All authors have reviewed and approved the final version of the manuscript. CRediT: Kai-Chun Chou: Conceptualization (lead), Data curation (lead), Investigation (lead), Methodology (lead), Project administration (supporting), Supervision (supporting), Visualization (lead), Writing–original draft (lead), Writing–review and editing (equal). David C. Zeitz: Data curation (supporting), Investigation (supporting), Methodology (supporting), Writing–original draft (supporting), Writing–review and editing (equal). Mariam Khvichia: Data curation (supporting), Methodology (supporting), Writing–original draft (supporting), Writing–review and editing (equal). Jin Z. Zhang: Conceptualization (supporting), Funding acquisition (lead), Project administration (lead), Resources (lead), Supervision (lead), Visualization (supporting), Writing–original draft (supporting), Writing–review and editing (equal).

The authors declare no competing financial interest.