We calculate the Gibbs potential $\ensuremath{\Gamma}(M)$ of a one-dimensional metal at constant magnetization
M to second order in the screened electron-electron interaction
U. At zero temperature we find that $\ensuremath{\Gamma}(M)$ contains nonanalytic corrections proportional to ${M}^{2}\phantom{\rule{0.2em}{0ex}}\mathrm{ln}\ensuremath{\mid}M\ensuremath{\mid}$ and ${\ensuremath{\mid}M\ensuremath{\mid}}^{3}$, implying that a possible paramagnetic-ferromagnetic quantum phase transition in one-dimensional metals must be first order.
A quantum model with
n-spin interactions is studied in the ground state. When $n\ensuremath{\rightarrow}\ensuremath{\infty}$ the model is solved exactly by mean-field theory and undergoes a first-order transition. The transition is expected to become second order for
n<nc. With exploitation of self-duality and suitably improved mean-field approximations, a recently proposed criterion leads to the conjecture
nc=3 in one dimension. For the two-dimensional case the transition appears to be definitely of first order for
n=3(2<nc<3).