Document Type


Date of Original Version



The linear, spin- ½ , alternating Heisenberg chain has attracted theoretical and experimental attention from physical chemists for about two decades, particularly in relation to spin exciton theory and the properties of linear, exchange-coupled free radicals. The model is somewhat unfamiliar to physicists but has become of increasing interest recently, primarily because of its relation to spin-Peierls transition systems. A striking feature of this model is that it has so far proved resistant to any form of analytic attack. Existing theories are therefore all approximate, and are not in agreement with one another. In particular, there is disagreement about the existence of an energy gap in the excitation spectrum for nonzero alternation, such a gap being crucial to spin-Peierls theory and spin exciton theory. In this paper we employ the method which has so far proved more reliable than any other approximate technique, namely the method of extrapolating exact finite-chain calculations to the thermodynamic limit. Our study is an extension of earlier work in this direction, and focuses on the nature of the ground state and on low-lying excitations in general, and the existence and properties of the gap in particular. We introduce the features of the linear alternating antiferromagnet through an initial description of the spin-Peierls transition and with brief reference to organic free radicals and spin exciton theory. This is followed by a survey of existing approximate theories. Features of the excitation spectrum are discussed and finite-chain extrapolations for the ground-state energy and energy gap as a function of alternation are presented. Comparisons are made with similar procedures performed on exactly solvable models, as a test of the expected accuracy of the extrapolations. Excitation spectra for a variety of other alternating models, classical and quantum, are calculated and surveyed comparatively. An unusual variety of behavior is observed, with striking differences between quantum and classical systems. Finally, a detailed comparison is made between our results and those of other approximate methods, including the new quantum renormalization-group approach. Particular attention is paid to values for the T = 0 spin-Peierls critical exponents.