Charge neutrality and an expected itinerant nature makes excitons potential transmitters of information. The creation in insulators of non-moving excitons (bound states from electrons and holes located at the minimum and maximum of the conduction band and valence band, repectively) by optical excitation is fairly standardand and has been widely studied both theoritically and experimentally.

Exciton mobility remains inaccessible to traditional optical experiments that only create and detect excitons with negligible momentum or group velocity. Exciting a moving bound state only with light involves a higher order process due to momentum conservation. The cross section of such processes scales with the interaction strength that is usually very weak. Thus, excitons with large non-zero group velocity have been rarely investigated.

 On the other hand, robust mobile excitons in metals have been elusive, as the screening effect usually suppresses their formation. The existence of excitonic bound states in three-dimensional metals requires the strength of the screened Coulomb ineraction to exceed a high threshold, which is usually not fulfilled (althought short-lived transient excitons tied to the surface of metal have been observed).

However, low-dimension (1D or 2D) usually enhance the Coulomb interaction effect, and even can separate charges with opposite sign (electrons and holes) in different layers and chains, which could enhance the life time of excitonic quasi-particles. Moreover, low density of states near Fermi level could decrease weaken the screening effect. In addition to low-dimension and low carrier density, exciton formation in metal is further enhanced if the involved quasiparticles are fairly heavy, which increases the binding erergy of the resulting exciton. The effective mass of dilute conduction electrons can be substantially increased by the coupling to relatively soft bosons (e.g., phonons), resulting in heavy quasiparticles close to the Fermi level.

As discussed above, athough the ingredients for the formation of mobile excitons are criticle, but all of the conditions seem to be present in the well-known Q1D material TaSe3 (Fig.1). Very recently, using angle-resolved photoemission spectroscopy, we detect dispersing excitons in quasi-one-dimensional metallic trichalcogenide TaSe3. The low density of conduction electrons and low dimensionality in TaSe3 combined with polaronic renormalization of the conduction band and a poorly screened interaction between these polarons and photo-induced valence holes leads to various excitonic bound states that we interpret as intrachain and interchain excitons, and possibly trions.

We have studied the metallic phase of TaSe3 using ARPES. We observe the following features at low temperature (Fig.2): (1) Several side-valence bands (SVBs) appear exclusively above a pronounced VB – in contrast to most ARPES spectra that report sidebands. (2) Their dispersions are roughly parallel to each other and to that of the VB. (3) When the surface carrier density is increased by doping, the energy separations between the SVBs increase. (4) Close to the Fermi level, the CB is heavily renormalized, the coherent quasiparticle peak following a W-shaped dispersion. As we will argue, observations (1-4) suggest that the SVBs result from strong coupling between the valence and conduction electrons and involve mobile bound states (excitons, and perhaps trions) that have not been observed so far using ARPES. Up to now, ARPES has detected excitonic physics only in the form of the consequences of excitonic condensation, as seen e.g., in the electronic structure of Ta2NiSe5 near the Fermi level, or through the band folding due to a finite momentum condensate, e.g., in 1T TiSe2.

As explained in Fig.3, the position, shape, and doping dependence of the SVBs can be best explained in terms of a moving bound state between the photo-hole with non-zero gorup velocity and large momentum in the VB of the given chain and one quasi-particle in the CB on the same or on neighboring chains. We propose the following interpretation: The MVB arises from the excitation of a single photo-hole in the VB, see Fig. 3cI. The top-most SVB above the MVB is the K-dependent threshold to a continuum consisting of an exciton and a free quasihole in the CB, the exciton being a moving bound state between a QP in the CB and a valence hole on the same chain, sharing the same group velocity Fig. 3cII. The SVBs closer to the MVB can be of two distinct origins. One possibility is that the particle-hole excitation in the CB is created on a chain neighboring the one hosting the valence hole, leading to an interchain exciton Fig. 3cIII, with lower binding energy than intrachain excitons. Alternatively, one can have thresholds to more complex continua, the simplest one consisting in a photo-hole accompanied by two particle-hole excitations on neighboring chains Fig. 3cIV, whereby both quasiparticles bind to the valence hole to form a mobile trion.

The material TaSe3 is a Q1D metal which combines a low density of conduction electrons with a polaronic renormalization of the low-energy quasiparticles. All these ingredients enhance the spectral weight in photoemission for composite excitations involving an exciton and a hole from the CB. Our experiments show that the excitons come with different internal structure, presumably depending on whether the involved holes and electrons belong to the same chain or neighboring ones, or whether the hole binds one or two conduction electrons (resulting in an exciton or a trion, respectively).

Interchain excitons are quasi-1D cousins of bilayer excitons in layered 2D materials, such as transition metal dichalcogenides. They are of particular interest as they may have a significantly longer life time than intra-chain excitons due to the spatial separation of the particle and the hole.

The work was published in Nature Materials on 22 Feb 2022: https://www.nature.com/articles/s41563-022-01201-9