It has been proposed in 1961 by Mott that a semimetal may be unstable towards an insulating ground state, when electrons and holes bind together through the Coulomb interaction and form excitons [1]. On this basis, it was elaborated a few years later that both a semimetal and a semiconductor can undergo this instability [2]. A phase transition occurs at low temperature, when the excitons condense in a macroscopic state, giving rise to the so-called excitonic insulator phase. Experimental observation of this phase has proven to be very challenging since its theoretical prediction and not much is known about the time-domain dynamics of this exotic phase.
In this talk, I will discuss the case of the semiconductor Ta₂NiSe₅, which has a direct band gap and displays a semiconductor-semiconductor phase transition at about 330 K [3]. Ta₂NiSe₅ has already been proposed for the realization of an excitonic insulator phase, based on angle-resolved photoemission spectroscopy (ARPES) data [3]. Its valence band at Γ shows a peculiar flattening at low temperature and is anomalously broad, explained as the consequence of strong excitonic fluctuations [4].
Here I will present time-resolved ARPES data of Ta₂NiSe₅. I will show how the band gap of Ta₂NiSe₅ can be controlled in a pump-probe experiment by varying the pump fluence. While for low fluences the band gap is transiently decreased, we observe that it can be increased for high fluences. I will argue that this observation is a direct consequence of the exciton condensate being trapped in a non-thermodynamical state where it is transiently strengthened.

References:
[1] N. Mott, Phil. Mag. 6, 287 (1961).
[2] D. Jérome et al, Phys. Rev. 158, 462 (1967).
[3] F.J. Di Salvo et al., J. Less Common Metals 116, 51 (1986).
[4] Y. Wakisaka et al., Phys. Rev. Lett. 103, 026402 (2009).
[5] K. Seki et al., Phys. Rev. B 90, 155116 (2014).
[6] S. Mor et al., submitted, arXiv:1608.05586.