Intermolecular vibrational energy transfer enabled by microcavity strong light–matter coupling

See allHide authors and affiliations

Science  08 May 2020:
Vol. 368, Issue 6491, pp. 665-667
DOI: 10.1126/science.aba3544

Atypical vibrational interactions

Vibrational energy transfer (VET) between solute molecules is generally unfavorable in liquids because of weak intermolecular forces. Xiang et al. measured the two-dimensional infrared spectrum of a molecular mixture, W(CO)6 and W(13CO)6, with saturated concentrations in a binary solvent embedded in an optical microcavity. This experiment showed that the VET between the asymmetric stretch vibrations of two solute molecules is enhanced via polaritonic intermediate states formed by a strong coupling with the cavity mode. The efficiency is modulated by the cavity lifetime, which provides an opportunity to control the VET process in the liquid phase. This could lead to various practical implementations.

Science, this issue p. 665


Selective vibrational energy transfer between molecules in the liquid phase, a difficult process hampered by weak intermolecular forces, is achieved through polaritons formed by strong coupling between cavity photon modes and donor and acceptor molecules. Using pump-probe and two-dimensional infrared spectroscopy, we found that the excitation of the upper polariton, which is composed mostly of donors, can efficiently relax to the acceptors within ~5 picoseconds. The energy-transfer efficiency can be further enhanced by increasing the cavity lifetime, suggesting that the energy transfer is a polaritonic process. This vibrational energy-transfer pathway opens doors for applications in remote chemistry, sensing mechanisms, and vibrational polariton condensation.

Vibrational energy transfer (VET) is ubiquitous for many molecular processes in the condensed phase, ranging from chemical catalysis (1, 2) to biological signal transduction and molecular recognition (3, 4). Due to through-bond anharmonic couplings, intramolecular and solute-solvent VET is widespread, leading to rapid intramolecular vibrational redistribution (IVR) that competes with other chemical events (5, 6). However, through-space, selective intermolecular (solute-solute) VET is relatively rare. The scarcity of intermolecular VET is a consequence of weak intermolecular forces. Relative to electronic transitions, which readily undergo intermolecular energy transfer [e.g., via the Förster and Dexter mechanisms (7, 8)], vibrational transition dipole moments are smaller by a factor of 10 to 100 (9), leading to uncompetitive intermolecular dipole-dipole couplings when compared with their electronic counterparts (Fig. 1A, top). As such, intermolecular VET is usually obfuscated by IVR.

Fig. 1 Strongly coupled system between W(CO)6 and W(13CO)6 in a hexane/DCM mixture and a cavity.

(A) Schematic illustration showing that VET between vibrational modes of W(CO)6 and W(13CO)6 molecules is unfavorable in solution (top) but is enabled by strong coupling of the molecular system to an infrared cavity mode (bottom). (B) Diagram of the 2D IR pulse sequence along with the IR spectrum and energy diagram of the system. (C) Transmission spectra of the polaritonic system as a function of incidence angle; white and green dashed lines denote bare W(CO)6 and W(13CO)6 vibrational transitions, respectively. (D) Hopfield coefficients for LP, MP, and UP as a function of incidence angle, calculated as described in supplementary materials section 11.

Here we report a state-of-the-art strategy to engineer intermolecular vibrational interactions via strong light–matter coupling. When a highly concentrated molecular sample is inserted into an optical microcavity [e.g., a Fabry-Perot (FP) cavity] or placed onto a plasmonic nanostructure (10), the confined electromagnetic modes interact reversibly with the collective macroscopic molecular vibrational polarization such that hybridized light–matter states, known as vibrational polaritons, are formed (916). Owing to the delocalized nature of polaritons, the relaxation kinetics of strongly coupled systems differs substantially from that of their weakly coupled counterparts (13). Although pioneering studies have demonstrated such effects in the context of electronic energy transfer (1721), intermolecular VET under strong light–matter coupling seems to operate by different mechanisms, as we describe below. Furthermore, given the scarcity of selective intermolecular VET in condensed phases, its polaritonic counterpart introduces a powerful concept that may alter the course of ground-state chemistry in solution (22).

To study cavity-assisted intermolecular VET, we designed a strongly coupled system composed of a microcavity and ensembles of two vibrational modes from different molecules. We encapsulated an equimolar solution of W(CO)6 and W(13CO)6 in hexane/dichloromethane (DCM) solvent (total concentration 105 mM, 1/1 volumetric ratio; see supplementary materials and methods for details) in a FP cavity. These molecules are ideal for achieving vibrational strong coupling, as they have degenerate asymmetric stretch modes with high oscillator strength and narrow linewidths. The cavity with thickness L has resonances at wavelengths λ=2nL, where n = 1,2,3, … is the cavity mode order. Because the carbonyl asymmetric stretches of W(CO)6 and W(13CO)6 absorb at 1980 and 1938 cm−1, respectively, a cavity with L = n × 2.5 μm has modes that are nearly resonant with both vibrational transitions. In our experiments, unless specifically noted, we kept L at 12.5 μm and strongly coupled the fifth-order cavity modes to the vibrations.

For each molecular subsystem, the light–matter coupling g is proportional to C, where C is the concentration of the absorbers. Given a large enough C, each molecular subsystem satisfies g > Γvib, Γcav, where Γvib and Γcav are the full widths at half maximum of the vibrational and cavity modes, respectively. Therefore, the vibrational and cavity modes (hereafter referred to as basis modes) hybridize and form new normal modes, denoted as upper, middle, and lower polaritons (UP, MP, and LP) (17, 19, 23) (Fig. 1B, bottom). Each polariton consists of a superposition of the basis modes. The polariton resonant frequency and composition, characterized by Hopfield coefficients, can be controlled by changing the incidence angle (Fig. 1, C and D). For example, at 15° incidence angle, the UP consists of 59.4% W(CO)6 carbonyl asymmetric stretch, 4.3% analogous vibration in W(13CO)6, and 36.3% cavity photon, whereas the LP is composed of 5.8, 60.8, and 33.4% of the respective basis modes. As discussed later, this information is essential to investigate the ability of strong coupling to mediate intermolecular VET.

We used two-dimensional infrared spectroscopy (2D IR) (1214) to show VET from W(CO)6 to W(13CO)6. In 2D IR, if the UP was pumped and VET occurred, a substantial population of W(13CO)6 excited states would be generated and the intensity of the corresponding cross peak would rise. In Fig. 2, we compare 2D IR spectra of the W(CO)6/W(13CO)6 mixture inside and outside the microcavity. The 2D IR spectrum of the bare W(CO)6/W(13CO)6 mixture (Fig. 2A) confirms the absence of energy transfer between vibrational modes. It shows two pairs of diagonal peaks, corresponding to excitations of asymmetric carbonyl modes of W(CO)6 and W(13CO)6, respectively, whose vibrational lifetimes are ~200 ps. There are no cross peaks (dashed black box in Fig. 2A), thus indicating the absence of intermolecular VET (supplementary materials section 2).

Fig. 2 Comparison of 2D IR spectra inside and outside of the microcavity.

2D IR spectra of (A) uncoupled and (B) strongly coupled W(CO)6/W(13CO)6 with a total concentration of 105 mM in binary solvent (hexane/DCM), along with the corresponding linear spectra of the two systems (top panels). The strongly coupled sample was taken at an incidence angle of 15° (Fig. 1D), where the cavity resonance is kept at 1961 cm−1. The dashed box in (A) indicates the absence of cross peaks. The red and black boxes in (B) indicate the [ωUP, ωLP] and [ωUP, ωMP] cross peaks, respectively.

The strongly coupled W(CO)6/W(13CO)6 system provides a substantially different picture. The 2D IR spectrum (Fig. 2B) shows several cross peaks at delay time t2 = 30 ps, indicating cavity-induced intermolecular correlations. At t2 = 30 ps (greater than the polariton lifetime), the remaining pumped energy equilibrates to the first excited state of dark modes, as determined in previous studies (12, 15). Furthermore, the W(13CO)6 dark modes have v=1v=2 transitions (v, vibrational state) at 1917 cm−1, whereas those of W(CO)6 are at 1961 cm−1. Thus, the transient absorptions at probe frequency ω3 = ωLP (~1920 cm−1) and ωMP (~1959 cm−1) provide an optical window into population dynamics of the W(13CO)6 and W(CO)6 reservoir modes, respectively (supplementary materials section 12). In particular, the cross peak at ω1 = ωUP and ω3 = ωLP (denoted as [ωUP, ωLP] hereafter; see red box in Fig. 2B) suggests that a larger-than-expected (on the basis of Hopfield coefficients) fraction of the energy in UP [dominated by W(CO)6] was transferred to the dark W(13CO)6 modes after 30 ps, a signature of intermolecular VET. We also conducted 2D IR experiments with the pump beam tailored to selectively excite |UP><UP| population states, and a similarly strong cross peak appeared at ω3 = ωLP. Thus, the [ωUP, ωLP] and [ωUP, ωMP] cross peaks (Fig. 2B) arise from a UP population, whereas pump-driven coherences such as |UP><LP| and |UP><MP| do not contribute to the observed spectral features (supplementary materials section 5).

We compared the cross-peak intensities at [ωUP, ωLP] (IUP,LP; red box in Fig. 2B) and [ωUP, ωMP] (IUP,MP; black box in Fig. 2B) to determine the equilibrated excited-state populations of W(13CO)6 and W(CO)6 (supplementary materials section 12) that arise from UP relaxation. On the basis of the Hopfield coefficients (for an incidence angle of 15°), we ascertained that only 4.3% of the UP energy would be stored in W(13CO)6 reservoir modes, whereas 59.4% would be allocated to W(CO)6 dark states. Therefore, in the absence of VET, the average population ratio between these reservoir modes (IUP,MP/IUP,LP) must be approximately equal to the ratio of the corresponding Hopfield coefficients: 59.4%/4.3% ≈ 14/1. However, experimentally, IUP,MP/IUP,LP is 2.5/1. This observation suggested that after the UP population was optically generated, its energy was preferentially channeled to W(13CO)6, increasing the relative peak intensity of [ωUP, ωLP]. To further examine this argument, we conducted similar measurements with a blue-detuned cavity mode so that UP had a greater fraction of W(CO)6 (supplementary materials section 7). Even at a Hopfield coefficient ratio of 25/1, there is still a substantial amount of energy transfer (IUP,MP/IUP,LP = 2.6 ± 0.1). Additional evidence supporting intermolecular VET is given by the anisotropy decay of the [ωUP, ωMP] and [ωUP, ωLP] peaks (supplementary materials section 8).

We then used pump-probe spectroscopy to monitor the VET dynamics when only the UP population was excited (Fig. 3A). The intensity of the [ωUP, ωLP] peak increased with a time constant of 5.7 ± 0.6 ps. This value represents the time scale for energy transfer from UP to W(13CO)6 reservoir modes. By contrast, the direct relaxation of UP to W(CO)6 happened much faster than VET, with a lifetime of 1.5 ± 0.3 ps, as indicated by the rising dynamics of IUP,MP. The decay of IUP,MP is composed of a fast and a slow component. The fast dynamics has a lifetime of 7 ± 2 ps, similar to the rising time of IUP,LP. Thus, it implies energy “leakage” from the W(CO)6 mode to the W(13CO)6 mode (see supplementary materials section 13). The slow component, whose decay extends beyond the time range of our scan, should correspond to the population relaxation of reservoir W(CO)6 (12, 15).

Fig. 3 Dynamics and cavity-thickness dependence of polariton-enabled intermolecular VET.

(A) Dynamics of [ωUP, ωLP] and [ωUP, ωUP] peak integrals and the fitting results. The sample was taken at an incidence angle of 15° (Fig. 1D). (B) Plot of IUP,MP/IUP,LP as a function of cavity thickness at t2 = 30 ps. Error bars represent the SD of three independent scans.

To confirm the importance of cavity modes in facilitating polariton VET, we attempted to enhance VET by increasing the cavity thickness L. In Fig. 3B, we present the IUP,MP/IUP,LP ratio for the same molecular mixture in cavities with L = 5, 12.5, and 25 μm, corresponding to 1.12-, 2.80-, and 5.60-ps cavity lifetimes, respectively (supplementary materials section 4). This ratio, which reflects the efficiency of VET at 30 ps, increased with increasing L. Because thicker cavities have longer lifetimes, this dependence suggests that a larger fraction of UP energy was collected in W(13CO)6 modes as polariton decay by photon leakage slowed. This property substantiates that intermolecular VET involves polaritonic intermediate states (supplementary materials section 13). The nature of the ultrafast energy redistribution process requires further study, which suggests that, in contrast to measurements performed on organic microcavities (17, 19), previously unexplored mechanisms dominate the relaxation kinetics reported here. Possible mechanisms include polariton-mediated scattering and interaction of polaritons with other dark modes.

The reported concept of polariton-enabled intermolecular VET could be expanded to the selective promotion or suppression of vibrational energy transport channels. The preferential relaxation to lower energy states is a key process for IR polariton condensation, remote energy transfer (17, 18), and cavity chemistry (22, 24).

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S19

Tables S1 and S2

References (2630)

References and Notes

Acknowledgments: The authors thank T. Porter from C. Kubiak’s group for his efforts in synthesizing the W(13CO)6 compounds. B.X. and W.X. thank J. Kim for insightful discussion on fluorescence resonance energy transfer. Funding: B.X. is supported by a Roger Tsien Fellowship. B.X., L.C., J.W., and W.X. acknowledge support from AFSOR YIP grant FA9550-17-1-0094. Z.Y and W.X. acknowledge support from NSF CAREER award DMR-1848215. The research instrument is supported by AFOSR DRUIP FA9550-18-1-0451. The development of the kinetic model was carried out by M.D. and J.Y.-Z. under funding support from the U.S. Department of Energy, Office of Science, Basic Energy Sciences, CPIMS Program under Early Career Research Program award DE-SC0019188. R.F.R. elucidated signatures of the energy-transfer processes in the nonlinear spectra with funding from Air Force Office of Scientific Research award FA9550-18-1-0289. Author contributions: W.X. and J.Y.-Z. conceived the original idea. W.X. supervised the overall research and J.Y.-Z. supervised the theoretical work. B.X. and W.X. designed the experiments. B.X., L.C., Z.Y., and J.W. conducted the experimental work. B.X. and W.X. analyzed experimental data. R.F.R., M.D., and J.Y.-Z. developed theoretical work. B.X., R.F.R., M.D., J.Y.-Z., and W.X. interpreted and discussed the experimental and theoretical results and wrote the final manuscript. Competing interests: The authors declare no competing interests. Data and materials availability: All data needed to support the conclusions of the main text and supplementary materials have been uploaded to Zenodo (25).

Stay Connected to Science

Navigate This Article