We study the role of “$\theta$ terms” in the action for three-dimensional $U\left(1\right)$ symmetric tensor gauge theories, describing quantum phases of matter hosting gapless higher-spin gauge modes and gapped subdimensional particle excitations, such as fractons. In Maxwell theory, the $\theta$ term is a total derivative which has no effect on the gapless photon, but has two important, closely related consequences: attaching electric charge to magnetic monopoles (the Witten effect) and leading to a Chern-Simons theory on the boundary. We will find that a similar story holds in the higher-spin $U\left(1\right)$ gauge theories. These theories admit generalized $\theta$ terms which have no effect on the gapless gauge mode, but which bind together electric and magnetic charges (both of which are generally subdimensional) in specific combinations, in a higher-spin manifestation of the Witten effect. We derive the corresponding Witten quantization condition. We find that, as in Maxwell theory, imposing time-reversal invariance restricts $\theta$ to certain discrete values. We also find that these new $\theta$ terms imply a nontrivial boundary structure. The boundaries host fracton excitations coupled to a tensor $U\left(1\right)$ gauge field with a Chern-Simons-like action, in both chiral and nonchiral varieties. These boundary theories open a door to the study of $U\left(1\right)$ fracton phases described by tensor Chern-Simons theories, not only on boundaries of three-dimensional systems, but also in strictly two spatial dimensions. We explicitly work through three examples of bulk and boundary theories, the principles of which can be readily extended to arbitrary higher-spin theories.

• Received 12 July 2017
• Revised 7 September 2017

DOI:https://doi.org/10.1103/PhysRevB.96.125151