Photoinduced changes in the cuprate electronic structure revealed by femtosecond time- and angle-resolved photoemission

The dressing of quasiparticles in solids is investigated by changes in the electronic structure $E$($k$) driven by femtosecond laser pulses. Employing time- and angle-resolved photoemission on an optimally doped cuprate above $T_{\mathrm{c}}$, we observe two effects with different characteristic temporal evolutions and, therefore, different microscopic origins. First, a marked change in the effective mass due to the 70-meV kink in $E$($k$) is found to occur during the experiment's 100-fs temporal resolution and is assigned to laser-driven perturbation of an electronic interaction dressing the bare band. Second, a change in $k_{\mathrm{F}}$ is explained by effective photodoping due to particle-hole asymmetry and offers opportunities for ultrafast optoelectronic switches based on an optically driven insulator-superconductor transition.

Figure 1

(a)–(c) Photoelectron intensity in a false color representation as a function of electron momentum $k-k_{\mathrm{F}}$ and binding energy with respect to $E_{\mathrm{F}}$ before, during, and after optical excitation, respectively, after applying a Lucy-Richardson deconvolution algorithm (Ref. [19]). The incident pump laser fluence was 300 μJ/cm${}^2$. Panel (d) shows the momentum-dependent intensity at $E_{\mathrm{F}}$ integrated within ±10 meV before and during optical excitation as well as the respective differences. Solid black lines are Lorentzian fits. MDC widths as a function of binding energy are shown in panel (e) for 300 μJ/cm${}^2$ fluence, $t$ = −5 ps (blue circles), and $t$ = 0 fs (red squares). The solid line $\left(\Delta k_{\Omega }\right)$ represents a linear least-squares fit to $E-E_{\mathrm{F}}<-70$ meV, $t$ = −5 ps MDC widths extrapolated to $E_{\mathrm{F}}$. The dashed line has the same slope and is shifted upwards by 6 meV. The inset shows $\Delta k_{\Omega }-\Delta k$ versus energy for both delay times to highlight the pump-induced change.

Figure 2

(a)–(d) Laser-pump-induced photoemission intensity difference $I\left(t\right)-I\left(t_0\right)$ as a function of momentum and energy in a blue-white-red false color representation at indicated delay times. (e)–(h) show the respective dispersions $E$($k$) determined from MDC maxima in red. In black, the dispersion before optical excitation at $t_0$ is shown for comparison. Dotted lines through $E_{\mathrm{F}}$, $k_{\mathrm{F}}$, and $E-E_{\mathrm{F}}=-70$ meV are guides to the eye.

Figure 3

Relative photoinduced change in the effective mass $m_{\%}^{*}$ for different optical excitation densities. The effective mass $m$*($t$) determined at time delay $t$ is normalized to $m$*($t_0$) before optical excitation. This practice eliminates ambiguity associated with the unknown “bare” noninteracting dispersion. The pink line indicates the pump-probe cross correlation of 100-fs temporal XC width.

Figure 4

Effective photoinduced hole doping δ$x$ (a) as a function of fluence Φ and (b) as a function of delay time after normalizing to the numbers of absorbed pump photons per unit cell. The analysis is based on the time-dependent shift in $k_{\mathrm{F}}$ (see Figs. 1 and 2). The determination of the number of holes per CuO${}_2$ unit cell shown in (a) assumes the model of the Fermi surface in cuprates by Yang et al. [25]. The lines representing least-squares linear fits to the fluence dependent δ$x$ assume a vanishing photodoping for an absent excitation. The points in (b) are derived from the slopes of these linear fits. The pink line in (b) guides the eye and is an exemplary single exponential decay function convoluted with the cross-correlation width for pump and probe laser pulses. Errors in $v_{\mathrm{F}}$ and $k_{\mathrm{F}}$ result from the least-squares fits.