Competing Trion and Exciton Dynamics in a Quasi-One-Dimensional Correlated Semiconductor

Abstract

Strong Coulomb interactions in low-dimensional quantum materials give rise to emergent bound states such as excitons and trions, which play a central role in correlated electronic phases. In quasi-one-dimensional systems, equilibrium photoemission studies have reported signatures of trions, suggesting an unusually robust state, as opposed to conventional semiconductors where trions typically appear only as excited states stabilized by carrier doping. Here, we show that optical excitation of undoped Ta2NiS5 - a correlated quasi-one-dimensional semiconductor - generates a pronounced and long-lived trion population, demonstrating that such states can be dynamically induced even in the absence of doping. Using time- and angle-resolved photoemission spectroscopy we track the dynamics of a bright, localized in-gap state that emerges following photoexcitation and identify it as a transient trion population. We uncover an unconventional trion formation pathway and a fluence-dependent competition between trions and excitons. These findings extend ultrafast quasiparticle photoemission spectroscopy to complex bound states in bulk quantum materials, enabling the dynamical control of charged and neutral excitations.

Figures (10)

• Figure 1: Time-resolved ARPES of Ta$_2$NiS$_5$(A) Time-resolved ARPES spectra of Ta$_2$NiS$_5$ oriented to $\Gamma-X$ direction and pumped with a $800$ nm ultrafast pulse ($160$$\mu$J/cm$^{2}$). The parabolic cutoff at the bottom of each panel is the photoemission horizon of the $6$ eV probe. (B) ARPES spectrum of Ta$_2$NiS$_5$ obtained with $43$ eV photons. The gray rectangle marks the momentum and energy regions shown in each panel of (A). (C) Sketch of the Brillouin zone of Ta$_2$NiS$_5$, the momentum cut shown in (A) is marked with a red line. (D) EDC as a function of pump-probe delay. The ARPES intensity of each delay was integrated over $k_{x}=[-0.1,0.1]$ Å$^{-1}$. (E) Intensity vs delay of the CB and the in-gap feature integrated over the momentum and energy ranges defined by the red and purple boxes in (A), respectively.
• Figure 2: Temporal evolution of the QPF and trion formation (A) Ta$_2$NiS$_5$ trARPES spectrum at $t=0$ (pump and probe overlap), obtained with a $400$ nm pump pulse, and normalized by the EDC integrated over $k_{x}=[-0.25,0.25]$ Å$^{-1}$. In addition to the QP-filled gap, the extended energy range encompasses the bottom and top conduction bands, represented by the solid and dashed black guidelines, respectively. (B) Schematic of a typical temporal signature of trARPES intensity and the temporal fitting parameters. (C-E) Temporal curve fitting results for every energy-momentum bin of the data set presented in Fig. \ref{['fig:1_snapshots']}(A), showing $I_{max}$, $t_{max}$, and $\tau$ respectively. A black guideline to the eye traces the bottom CB. (F) schematic sketch of the optically driven generation of two types of QP: excitons and trions, and their photoemission signature. The pump pulse (NIR/VIS) excites an electron to the CB, where it forms a trion (upper branch) or exciton (bottom branch). Upon photoemission, the probe pulse (UV) breaks both types of quasiparticles, granting different kinetic energies to the emitted electrons according to their respective binding energies.
• Figure 3: QPF fluence dependence and time-dependent composition (A) trARPES snapshots at $t=300$ fs for three pump fluences ($800$ nm): $490$, $160$ and $50$$\mu$J/cm$^{2}$, respectively. The color scale is set to the minimum and maximum of each panel separately. (B) Temporal intensity curves of the three fluences, integrated over the red box in panel (A-i). The counts at negative delays were subtracted for each curve, and all were normalized to their peak. (C) QPF integrated intensity and the total quasiparticle population, $n_{X}+n_{T}$, from the rate equation results shown in black and gray, respectively. Exciton ($n_X$) and trion ($n_T$) populations in green and purple, respectively. (D) portion of excitons and trions of the total quasiparticle population $n_{X,T}$/$(n_{X}+n_{T})$. (E) sketch of the processes described by the rate equations. The two-particle processes (exciton formation and trion decomposition) are marked with an additional gray arrow, indicating the interaction with a photo-induced hole.
• Figure 4: Dynamics of QPF energy distribution (A) Sketch of the exciton and trion energy ladder with respect to the CBM. (B) EDC as a function of delay, identical to Fig. \ref{['fig:1_snapshots']}(D), overlaid with $E_{0}$ (red) and $E_{0}\pm\sigma$ (magenta). (C-E) amplitude $A$, width $\sigma$, and center energy $E_{0}$ extracted from fitting the QPF to a Gaussian (details in SM) for each delay of the middle fluence data set. Values for $t=0$, 25 fs were omitted due to unreliable fit. The gray line in D (E) shows the fit result to a decaying exponent (double exponent). (F-H) zoomed in view of (C-E) at small delays.
• Figure S1: Spectral properties of the QPF (A) Fermi surface (FS) map of the QPF, taken at $t=400$ fs using the deflection mode of the electron analyzer (without rotating the sample) and a sketch of the captured portion of the Brillouin zone. (B-C) Effective mass of the QFP at $t=800$ fs, EDCs (gray) were fitted with the sum of a Gaussian and an exponent (red) methods, and the peaks, marked in black, were extracted. (C) The positions of the peaks were plotted (black markers) on the spectrum and fitted with a parabola (black line), yielding the effective mass. (A) was obtained with $800$ nm pump pulse. (B-C) are extracted from the same data set as Fig. \ref{['fig:1_snapshots']} in the main text.