Non-2kF charge density wave induced by phonon dispersion in one-dimensional Peierls conductors

Abstract

Standard Peierls theory predicts that the wave-vector of the charge density wave (CDW) induced by the electron-phonon coupling in the ground-state of a one-dimensional conductor is twice the Fermi wave-vector 2kF. At moderately large electron-phonon coupling, beyond the TBA (transition by breaking of analyticity), it is known that this CDW becomes an array of interacting bipolarons. In the simplest models, such as the 1D adiabatic Holstein model, the bipolaron interactions are repulsive and the ground state remains a 2kF CDW of equidistant bipolarons. We show that (apparently) minor changes in the phonon spectrum of the model induce extra elastic forces between the bipolarons, which could break the bipolaronic 2kF CDW ordering. This effect is studied in detail in the modified 1D Holstein model, where the optical phonon has a non-zero dispersion which is chosen positive in order that an extra force that appears between the bipolarons be attractive. The Coulomb forces between the bipolarons are initially neglected. An accurate numerical study modelled on the standard analysis of multiphase points confirms that, in a large part of the phase diagram, the CDW ground state separates in two phases, which correspond to two CDW with different 2kF wave-vectors (which are generally commensurate) and thus different electronic densities. These separations, into phases called either 'parent' or 'non-parent', are studied on the basis of the Farey construction of rational numbers. The long-range Coulomb interaction between the charged bipolarons, forbids any macroscopic phase separation in two phases with different electronic densities. The bipolaron structure then has to be a periodic sequence of alternate domains of the two CDW phases on the microscopic scale. The resulting structure is a new CDW with a modulation wave-vector that is not 2kF. We suggest that the aberrant wave-vector observed in the real CDW system (TaSe4)2I could be interpreted along the ideas developed in this paper. Although this compound shares many of the usual properties of quasi-one-dimensional conductors with CDW, the component of the modulation wave-vector 0.085c* in the chain direction is, unusually, quite different from the value 2kF=c* expected from band filling. The CDW in (TaSe4)2I could correspond to a nonstandard bipolaron ordering due to a strong attractive elastic interaction between the bipolarons along the chain competing with the long-range Coulomb forces.