Two-Dimensional Impulsively Stimulated Resonant Raman Spectroscopy of Molecular Excited States

Monitoring the interactions between electronic and vibrational degrees of freedom in molecules is critical to our understanding of their structural dynamics. This is typically hampered by the lack of spectroscopic probes able to detect different energy scales with high temporal and frequency resolution. Coherent Raman spectroscopy can combine the capabilities of multidimensional spectroscopy with structural sensitivity at ultrafast timescales. Here, we develop a three-color-based 2D impulsive stimulated Raman technique that can selectively probe vibrational mode couplings between different active sites in molecules by taking advantage of resonance Raman enhancement. Three temporally delayed pulses generate nuclear wave packets whose evolution reports on the underlying potential energy surface, which we decipher using a diagrammatic approach enabling us to assign the origin of the spectroscopic signatures. We benchmark the method by revealing vibronic couplings in the ultrafast dynamics following photoexcitation of the green fluorescent protein.

Popular Summary

Light-induced processes in molecules rely on the efficient and directed conversion of photon energy into electronic and atomic motion. This conversion is tightly controlled by the underlying multidimensional energy surfaces, which describe how the potential energy of the system changes with vibrational and electronic configuration. Mapping these surfaces over multiple vibrational dimensions discloses the ultrafast evolution of the system but is typically hampered by the need for spectroscopic probes detecting different energy scales with high temporal and frequency resolution. Here, we introduce a 2D coherent Raman scheme to probe vibrational correlations pertaining to targeted electronically excited states.

Specifically, the technique allows one to coherently generate and track excited-state vibrational wave packets by means of three temporally delayed femtosecond pulses. The evolution of these wave packets is determined by the shapes of the vibrationally structured potential energy surfaces of the system. We use a diagrammatic approach to assign the origin of the different spectral features and map the multidimensional energy surfaces involved in the process. In particular, we demonstrate our approach by examining the first steps of the photoinduced dynamics in the green fluorescent protein—an efficient biomarker, first isolated in jellyfish, that glows green when exposed to blue light—revealing the vibronic couplings in the excited state.

Our technique provides the chance to directly access the structural conformation on the state in which the dynamics originate, disclosing the different stages of the reaction and improving our understanding of excited states in general.

Figure 1

Concept of 2D-ISRS. (a) Pulse sequence used for the experiment. (b) Energy ladder scheme illustrating an example of signal generation. Horizontal black lines represent the vibrational levels of the sample organized in three electronic manifolds. Dotted and solid arrows indicate light interactions with the ket and the bra side of the density matrix representing the state of the system. (c)–(e) Overview of the data analysis required in 2D-ISRS. (c) Transient absorption signals recorded along ${\mathrm{T}}_2$ for a given ${\mathrm{T}}_1$ over all probe wavelengths ${\lambda }_s$, and the vibrational coherence contribution, which is isolated by fitting and subtracting the electronic background. (d) At each ${\lambda }_s$, we collect the coherence for each ${\mathrm{T}}_1$ to build $S(T_1,T_2)$, which is (e) subsequently 2D Fourier transformed to yield $S({\mathrm{\Omega }}_1,{\mathrm{\Omega }}_2)$, after averaging the WLC probe wavelengths ${\lambda }_s$ along the desired resonance region.

Figure 2

Experimental characterization of wild-type GFP. (a) Photodynamics of GFP. Photoexcitation of the neutral ${\mathrm{A}}_0$ chromophore at 397 nm prepares A*, which decays with biphasic dynamics (2–3 and 8–10 ps) to the fluorescently active deprotonated I* chromophore [59, 60]. (b) Normalized absorption spectrum (gray) and transient absorption spectrum at 1 ps (green) and employed actinic and Raman pulse spectra (blue and orange). (c) Comparison of FC spectrum (blue) and A* Fourier amplitude spectrum binned in ${\mathrm{T}}_1$ to 300 fs to average out any effect of oscillatory modulations along ${\mathrm{T}}_1$ (orange). The absolute amplitude of the FC spectrum was scaled by 0.2 for clarity. Raman spectra were averaged over a probe wavelength region from 575 to 615 nm (stimulated emission), and coherent oscillations were Fourier transformed over a time delay of 1 ps (see Appendix pp3 for further details).

Figure 3

The 2D-ISRS characterization of the first excited electronic state of GFP. (a) Experimental 2D-ISRS map and (b) fit to the linearly displaced harmonic model. Vertical dotted lines correspond to frequencies of the GFP excited-state vibrations, extracted by fitting the principal diagonal. The blue dashed and blue solid arrows highlight combination bands involving the fundamental and overtone frequencies of the $1147\text{−}{\mathrm{cm}}^{-1}$ mode, respectively. As evidenced by the experimental map in panel (a), several couplings with the overtone are present, while there are no peaks on the blue dashed arrow. The red arrow highlights combination bands involving the mode at $1010{\mathrm{cm}}^{-1}$ that is absent along the principal diagonal. The green dashed arrow highlights a combination band between the 1147- and $112\text{−}{\mathrm{cm}}^{-1}$ modes. The dashed area around the principal diagonal has been scaled by 0.4 for visual purposes.

Figure 4

Theoretical interpretation of 2D-ISRS signals in the harmonic model. (a) Double-sided Feynman diagrams describing the contributions to the signal. (b) Different energy landscapes along the vibrational coordinate ${\mathrm{Q}}_i$ probed by the actinic pulse (left panel) and Raman and probe pair (right panel). In the harmonic model, even if three electronic energy potentials are involved, three different displacements among them have to be considered, since $d_1$ can be different from $d_3$, due to the relaxation of the molecule out of the FC region. (c) Simulation of the linearly displaced harmonic model for a two-mode system ${\omega }_h=730{\mathrm{cm}}^{-1}$ and ${\omega }_l=250{\mathrm{cm}}^{-1}$ considering three different choices of the displacement set, in which two of the six parameters are set equal to zero. For each panel, the two nondisplaced coordinates are indicated in gray by the color scheme on top. The left panel shows diagonal peaks obtained for $d_2^h$, $d_2^l=0$ and the other parameters different from zero; the central panel shows two cross peaks at ${\mathrm{\Omega }}_1=\pm {\omega }_l$, ${\mathrm{\Omega }}_2={\omega }_h$ obtained by switching off $d_1^h$ and $d_3^l$; in the right panel, the only vanishing displacements are $d_2^l$ and $d_3^h$, which lead to a diagonal peak at ${\omega }_l$ and two combination bands at ${\mathrm{\Omega }}_1={\omega }_h\pm {\omega }_l$ and ${\mathrm{\Omega }}_2={\omega }_l$. Red (blue) horizontal and vertical lines highlight the ${\omega }_l$ (${\omega }_h$) frequency.

Figure 5

Effect of mode mixing in 2D-ISRS for a harmonic model. The experimental spectra of GFP (a) are compared with the fit (b) obtained by considering the modes at 80, 822, 888, and $1147{\mathrm{cm}}^{-1}$, which are those relevant for the off-diagonal peaks at ${\mathrm{\Omega }}_1={\mathrm{\Omega }}_2-2294{\mathrm{cm}}^{-1}$. These features, not captured by the LDHO model, are explained by a Duschinsky rotation of ${\mathrm{\Theta }}_{\mathrm{D}}=90°$ between the 1147- and $80\text{−}{\mathrm{cm}}^{-1}$ modes, and they are highlighted by the rainbow color bar in the experimental map. The gray scale has been used for the peaks not relevant for these processes and already assigned by the LDHO model. In the bottom panels, the effect of the Duschinsky rotation is dissected by considering the minimal scenario of a single mode ($1147{\mathrm{cm}}^{-1}$), coupled to a nondisplaced low-frequency mode ($80{\mathrm{cm}}^{-1}$) and detected on a “spectator” mode ($822{\mathrm{cm}}^{-1}$). For ${\mathrm{\Theta }}_{\mathrm{D}}=0°$ (a,b), the system is described by a linearly displaced harmonic model, and a displacement $d_2^{1147}\ne 0$ causes the combination bands involving this mode and its overtones. The blue and cyan arrows indicate the position of combination bands associated with $1147{\mathrm{cm}}^{-1}$ and $2\times 1147{\mathrm{cm}}^{-1}$, respectively. (c,d) In the presence of a Duschinsky rotation ${\mathrm{\Theta }}_{\mathrm{D}}=90°$ between the projections of ${\mathrm{S}}_1$ and ${\mathrm{S}}_n$ along the normal coordinates of the 1147- and $80\text{−}{\mathrm{cm}}^{-1}$ modes, the combination band associated with the fundamental of the $1147{\mathrm{cm}}^{-1}$ is suppressed.