Photomolecular High-Temperature Superconductivity

The properties of organic conductors are often tuned by the application of chemical or external pressure, which change orbital overlaps and electronic bandwidths while leaving the molecular building blocks virtually unperturbed. Here, we show that, unlike any other method, light can be used to manipulate the local electronic properties at the molecular sites, giving rise to new emergent properties. Targeted molecular excitations in the charge-transfer salt $\kappa \text{−}{(\mathrm{BEDT}-\mathrm{TTF})}_2\mathrm{Cu}[\mathrm{N}{(\mathrm{CN})}_2]\mathrm{Br}$ induce a colossal increase in carrier mobility and the opening of a superconducting optical gap. Both features track the density of quasiparticles of the equilibrium metal and can be observed up to a characteristic coherence temperature $T^{*}\simeq 50\mathrm{K}$, far higher than the equilibrium transition temperature $T_{\mathit{C}}=12.5\mathrm{K}$. Notably, the large optical gap achieved by photoexcitation is not observed in the equilibrium superconductor, pointing to a light-induced state that is different from that obtained by cooling. First-principles calculations and model Hamiltonian dynamics predict a transient state with long-range pairing correlations, providing a possible physical scenario for photomolecular superconductivity.

Popular Summary

High-temperature superconductivity is found in a wide variety of organic conductors—synthetic metals composed of molecular building blocks whose nature and arrangement fully determine the material’s properties. Typically, these properties are altered by the use of chemical substitutions or by the application of external pressure. These approaches affect only the mutual arrangement of the molecular building blocks, leaving their intrinsic nature unchanged. Here, we use ultrafast midinfrared laser pulses to directly manipulate the molecular building blocks and induce transient superconductivity at a temperature much higher than that found at equilibrium.

In our experiment, we focus on the organic superconductor $\kappa \text{−}{(\mathrm{BEDT}\text{−}\mathrm{TTF})}_2\mathrm{C}\mathrm{u}\mathrm{[}{\mathrm{N}\mathrm{(}\mathrm{C}\mathrm{N}\mathrm{)}}_{\mathrm{2}}\mathrm{]}\mathrm{B}\mathrm{r}$. This compound has a superconducting transition temperature of 12.5 K, the highest for this family of superconductors. Using femtosecond midinfrared laser pulses tuned to excite specific molecular vibrational modes, we induce transient superconductivity at temperatures as high as 50 K. This change in behavior arises from how the laser pulses modulate the local molecular wave function and with it the electronic interactions in the solid.

This discovery provides a possible physical scenario for photomolecular superconductivity and a new way of tuning electronic ground states in solids.

Figure 1

(a) Left: Crystal structure of $\kappa \text{−}{(\mathrm{BEDT}-\mathrm{TTF})}_2\mathrm{Cu}[\mathrm{N}{(\mathrm{CN})}_2]\mathrm{Br}$. Alternating blocks of conducting and insulating layers are formed out of donor BEDT-TTF (abbreviated as ET) and polymeric chain-forming acceptor $\mathrm{Cu}[\mathrm{N}{(\mathrm{CN})}_2]\mathrm{Br}$ molecules, respectively. The ET molecules are paired in dimers. Right: View of the pressure-dependent arrangement of the dimers in the $\kappa$-phase (visualized along the long axis of the molecules, i.e., top view from left panel). As (external or chemical) pressure is applied, the spacing between dimers changes, thus acting on the transfer integrals ($t$, $t^{\prime }$) and tuning the ground state of the material. (b) Left: Lennard-Jones potential, along with highest occupied molecular orbital (HOMO) plots for the first two vibrational levels. Violet and yellow colors denote the sign of the wave function. Right: Time evolution of the HOMO after pulsed excitation, creating a mixed population of $\nu =1$ and $\nu =0$ states, and the corresponding arrangement of the ET dimers (seen from the top). The periodic changes in the wave function introduce modulations in the on-dimer Hubbard $U$ and transfer integral $t$, opening new dynamic pathways for controlling the material properties.

Figure 2

(a) Upper panel: Out-of-plane equilibrium optical conductivity of $\kappa \text{−}{(\mathrm{ET})}_2\mathrm{Cu}[\mathrm{N}{(\mathrm{CN})}_2]\mathrm{Br}$ measured at $T=15\mathrm{K}$ [14]. Lower panel: Midinfrared pump spectra tuned close to resonance with the ${\nu }_{27}$ molecular vibration. The eigenvectors corresponding to this mode are displayed in the upper panel [17]. (b) Upper panel: In-plane equilibrium optical conductivity measured at different temperatures between 15 K and 80 K. Lower panel: Frequency spectrum of the THz probe pulses used in our experiment.

Figure 3

In-plane reflectivity, real and imaginary parts of the optical conductivity measured in $\kappa \text{−}{(\mathrm{ET})}_2\mathrm{Cu}[\mathrm{N}{(\mathrm{CN})}_2]\mathrm{Br}$ at equilibrium (red circles) and at $\tau \simeq 1\mathrm{ps}$ time delay after vibrational excitation (blue circles), at four different base temperatures between 15 K and 70 K. Lines are corresponding fits to the spectra. These fits were performed with a Drude-Lorentz model for the equilibrium response at all temperatures and for the transient spectra at $T=70\mathrm{K}$. A Mattis-Bardeen model for superconductors was used instead for the out-of-equilibrium response at $T\le 50\mathrm{K}$. All data have been taken upon vibrational excitation close to resonance with the $\mathrm{C}═\mathrm{C}$ stretching mode (${\nu }_{\mathrm{pump}}\simeq 1250{\mathrm{cm}}^{-1}$), with a pump fluence of $2\mathrm{mJ}/{\mathrm{cm}}^2$.

Figure 4

(a) Photoinduced optical gap $2\mathrm{\Delta }$, extracted from Mattis-Bardeen fits to the transient optical properties of Fig. 3, as a function of base temperature. (b) Blue, left scale: Temperature dependence of the effective number of “condensed” carriers per unit cell in the transient state, $N_{\mathrm{eff}}^{\mathrm{Trans}}$. Red, right scale: Corresponding effective number of mobile carriers per unit cell in the equilibrium metallic state before photoexcitation, $N_{\mathrm{eff}}^{\mathrm{Equil}}$ (see discussion in the text). These data have been taken upon vibrational excitation with a pump fluence of $2\mathrm{mJ}/{\mathrm{cm}}^2$.

Figure 5

(a) Dynamical evolution of the transient spectral weight in the real part of the optical conductivity as a function of pump-probe time delay. (b) Time evolution of the effective number of “condensed” carriers per unit cell in the transient state, $N_{\mathrm{eff}}^{\mathrm{Trans}}$. (c) Time evolution of the zero-frequency extrapolation of the optical conductivity, ${\sigma }_0={{\sigma }_1(\omega )|}_{\omega \to 0}$, extracted from Drude-Lorentz fits of the transient optical properties. Data have been taken at $T=30\mathrm{K}$ after vibrational excitation with a pump fluence of $2\mathrm{mJ}/{\mathrm{cm}}^2$.

Figure 6

Left panels: In-plane complex optical conductivity of $\kappa \text{−}{(\mathrm{ET})}_2\mathrm{Cu}[\mathrm{N}{(\mathrm{CN})}_2]\mathrm{Br}$ measured at $T=50\mathrm{K}$ in equilibrium (red circles) and at $\tau \simeq 1\mathrm{ps}$ time delay after photoexcitation (blue circles) at four different pump frequencies: ${\nu }_{\mathrm{pump}}=910{\mathrm{cm}}^{-1}({\lambda }_{\mathrm{pump}}=11\mu \mathrm{m})$, ${\nu }_{\mathrm{pump}}=1250{\mathrm{cm}}^{-1}({\lambda }_{\mathrm{pump}}=8\mu \mathrm{m})$, ${\nu }_{\mathrm{pump}}=1470{\mathrm{cm}}^{-1}({\lambda }_{\mathrm{pump}}=6.8\mu \mathrm{m})$, and ${\nu }_{\mathrm{pump}}=2000{\mathrm{cm}}^{-1}({\lambda }_{\mathrm{pump}}=5\mu \mathrm{m})$. Lines are corresponding fits performed with a Drude-Lorentz model. The ${\nu }_{\mathrm{pump}}=2000{\mathrm{cm}}^{-1}$ spectra display an enhancement of ${\sigma }_2(\omega )$ in the absence of a gap opening in ${\sigma }_1(\omega )$. This is indicative of a slight increase in metallicity, with no superconducting-like response. All data have been taken with a constant pump fluence of $2\mathrm{mJ}/{\mathrm{cm}}^2$. Right panel: Zero-frequency extrapolation of the transient optical conductivity, ${\sigma }_0={{\sigma }_1(\omega )|}_{\omega \to 0}$ (see main text), extracted from the spectra on the left for different pump wavelengths (blue circles, left scale). A data point at ${\nu }_{\mathrm{pump}}=1100{\mathrm{cm}}^{-1}({\lambda }_{\mathrm{pump}}=9\mu \mathrm{m})$ is also reported, for which the transient spectra look identical to those at ${\nu }_{\mathrm{pump}}=910{\mathrm{cm}}^{-1}$ (negligible pump-induced effect). The equilibrium out-of-plane imaginary dielectric function [14] is also displayed (black line, right scale) to visualize the resonant photoconductivity behavior on the $\mathrm{C}═\mathrm{C}$ ET vibration.

Figure 7

Left panels: Atomic motions corresponding to the $1470{\mathrm{cm}}^{-1}$ (a) and $920{\mathrm{cm}}^{-1}$ (b) modes of the ET molecule. Center panels: Time-dependent displacement along the normal coordinates and induced modulations of the on-site Hubbard $U$ interaction and hopping matrix element $t$, calculated with ab initio simulations in the frozen-phonon approximation for the two different driven modes. Right panels: Corresponding spatial evolution of the pair correlation function, calculated with a time-dependent triangular ladder Fermi-Hubbard model in the undriven case (blue dashed line) and driven case (blue solid lines) for the two modes. The insets display a schematic representation of the correlation between pairs on different sites.