The free energy per lattice site of a quantum spin chain in the thermodynamic limit
is determined by a single “dominant” eigenvalue of an associated quantum transfer
matrix in the infinite Trotter number limit. For integrable quantum spin chains
constructed from solutions of the Yang-Baxter equation, the quantum transfer
matrix may be taken as the transfer matrix of an inhomogeneous variant of
the underlying vertex model. Its spectrum can then be studied by Bethe
Ansatz methods and may exhibit universal features such as the emergence
of a conformal subspectrum in the low-temperature regime. Access to the
full spectrum of the quantum transfer matrix enables the construction of
thermal form factor series representations of the correlation functions of local
operators for the spin chain under consideration. These are claims, made by
physicists, whose rigorous mathematical justification sets up a long-term
research programme. In this work we implement first steps of this programme
with the example of the XXZ quantum spin chain in the antiferromagnetic
massless parameter regime and in the low-temperature limit. We rigorously
establish the existence and uniqueness of the solutions to a set of nonlinear
integral equations, that are equivalent to the Bethe Ansatz equations for
the quantum transfer matrix of this model, and explicitly characterise the
low-temperature form of these solutions. This allows us to describe that
part of the quantum transfer matrix spectrum that is related to the Bethe
Ansatz and that does not collapse to zero in the infinite Trotter number
limit. Within the considered part of the spectrum we rigorously identify
the unique eigenvalue of largest modulus and show that those correlations
lengths that diverge in the low-temperature limit are, to the leading order in
temperature, in one-to-one correspondence with the spectrum of the free boson
conformal field theory. Based on two conjectures, that are accepted in the physics
literature, but that could so far only be established in the opposite limit of high
temperatures, we prove that the eigenvalue of largest modulus in the subspectrum we
focus on corresponds, in fact, to the dominant eigenvalue. Its first-order term
in temperature is of a universal form conjectured long ago in the physics
literature.