In semiconductors, photoemission from photodoped carriers is straightforward: free carriers produce spectral weight at the band edge, and excitons create replicas of the valence band inside the gap. The features map directly onto the quasiparticle picture. But Mott insulators are not semiconductors — their elementary excitations are fractionalized into spinons, holons, and doublons, not conventional electrons and holes.
This paper shows that photoemission from photodoped one-dimensional Mott insulators is qualitatively different. When photodoped carriers (doublons and holons) remain unbound, the photoemission spectrum directly reflects the spinon dispersion — magnetic excitations, not charge excitations. The charge carriers are invisible; their magnetic counterparts dominate the spectrum.
When a doublon and holon bind into an exciton, replica structures of the lower Hubbard band appear inside the Mott gap, carrying contributions from both spinon and holon excitations. The distribution of this in-gap signal depends sensitively on how tightly the doublon-holon pair is bound. Photoemission from a Mott insulator exciton reveals the binding strength of constituents that are themselves fractionalized. The spectroscopic signature contains information about magnetic properties and carrier binding that has no analogue in the semiconductor case.
Correction (2026-11-05): I checked this essay against the raw arXiv abstract of 'Photoemission Signatures of Photoinduced Carriers and Excitons in One-Dimensional Mott Insulators' (arXiv 2603.18982). The abstract opens with 'We theoretically study', so these are calculated spectra, not measurements, and the essay does not say so. The abstract says that when doublons and holons are unbound the photoemission spectrum directly reflects the spinon dispersion, and that when they bind into an exciton, replicas of the lower Hubbard band appear inside the gap with contributions from spinons and holons, whose distribution depends sensitively on the degree of binding. My wording that the charge carriers are invisible and that magnetic counterparts dominate the spectrum is stronger than that, and the line that the signal has no analogue in the semiconductor case is my extension of the abstract's 'qualitatively different'.