A polariton can be seen as a molecule of light and matter. Much like an atom binds an electron to a nucleus, a polariton binds a photon (a particle of light) to a material excitation (a particle of matter), typically an exciton. There are various types of polaritons, depending on what the photon couples to, but the most popular type is the exciton-polariton. This also provides the basic picture which remains valid for most other types of polaritons.
The exciton is unstable, like a positron (it is in fact a bound electron-hole pair), an after a while, recombines to create another photon. The process repeats itself, for as long as the photon stays in the cavity. According to Quantum Mechanics, this generates a quantum superposition of the two possible states, just like the Schrödinger cat, but instead of being alive and dead simultaneously, the particle is light and matter simultaneously:
$$\ket{\mathrm{photon}}+\ket{\mathrm{exciton}}$$
These half-light/half-matter particles have fantastic properties, which they inherit from their underlying components:
They have a mass, like matter, but a very light-one (the photon has none in vacuum and a small effective one when confined).
Their dynamics consist of characteristic Rabi oscillations.[1] In presence of a weak nonlinearity, this turns into the problem of Josephson oscillations.[2]
The concept was first proposed theoretically, and christened so, by John Hopfield[3], but its most fruitful implementation, in 2D—the cavity-polariton—was discovered experimentally by Claude Weisbuch during a sabbatical stay in Arakawa's group in Tokyo.[4] He referred to it as his «Japanese effect.»[5] Weisbuch did not initially recognized it as a Hopfield polariton, or "bulk polariton", for which the photon is delocalized in the full 3D crystal, but as a cavity QED effect. Weisbuch was in fact already a polariton expert, having reported its first resonant observation 15 years earlier[6]. He didn't mention either surface polaritons or the closely related waveguide polaritons of K. Ogawa et al.[7]
The situation was quickly settled during the July (1993) Erice Summer School on "Confined Electrons and Photons: New Physics and Applications" which featured «heated sessions (involving in particular the two Elis, Eli Burstein and Eli Yablonovitch) on the nature of these excitations» according to Weisbuch himself.[8] The name of "Cavity-Polariton" was then agreed to well describe Hopfield's counterpart in reduced dimensionality, and be more suitable than Weisbuch's initial choice for merely Rabi splitting:
The term "vacuum field Rabi splitting" has so far been used for semiconductor microcavities in analogy to atomic physics where this effect was first observed. From a solid state physics point of view, where dispersion has to be considered, the term "cavity-polariton" is more appropriate.
The first appearance of the "cavity polariton" term was in Ref. [9], where the idea of the underlying dispersion was being formed. The breakthrough came in Ref. [10] where the now famous polariton dispersion was provided for the first time:
It seems that the theorists who should have predicted cavity-polaritons are C. Andreani et al.,[11] who narrowly missed it by overlooking the dimensionality mismatch could be fixed with a cavity. He seems, however, to have readily understood (and explained) it[12] to R. Houdré et al. as they were shaping the field.[5]
Important concepts (I'm particularly interested in) include:
A historical overview of polaritons before the cavity, from Hopfield's lone-warrior paper[3] that gets all the credit despite fairly serious variations with the finally adopted results (including in its operational definition, involving four operators as not taking the rotating wave approximation), along with other contributions.[13][14] The same can be then done from the experimental point of view, especially concerning cavity polaritons and their connections (or missed ones) to surface- and waveguide-polaritons.
The topic is covered at depth in our book Microcavities. Reviews include [15][16][17][18].
Historical overviews includes Refs. [19][5], etc.