Size Exponents of Branched Polymers/ Extension of the Isaacson-Lubensky Formula and the Application to Lattice Trees

Kazumi Suematsu,Haruo Ogura,Seiichi Inayama,Toshihiko Okamoto
DOI: https://doi.org/10.48550/arXiv.2203.01669
2022-03-03
Soft Condensed Matter
Abstract:Branched polymers can be classified into two categories that obey the different formulae: \begin{equation} \nu= \begin{cases} \hspace{1mm}\displaystyle\frac{2(1+\nu_{0})}{d+2} & \hspace{3mm}\mbox{for polymers with}\hspace{2mm}\displaystyle\nu_{0}\ge\frac{1}{d+1}\hspace{10mm}\text{(I)}\\[3mm] \hspace{5mm}2\nu_{0}& \hspace{3mm}\mbox{for polymers with}\hspace{2mm}\displaystyle\nu_{0}\le\frac{1}{d+1}\hspace{10mm}\text{(II)} \end{cases}\notag \end{equation} for the dilution limit in good solvents. The category II covers the exceptional polymers having fully expanded configurations. On the basis of these equalities, we discuss the size exponents of the nested architectures and the lattice trees. In particular, we compare our preceding result, $\nu_{d=2}=1/2$, for the $z$=2 polymer having $\nu_{0}=1/4$ with the numerical result, $\nu_{d=2}\doteq 0.64115$, for the lattice trees generated on the 2-dimensional lattice. Our conjecture is that while both the conclusions in polymer physics and condensed matter physics are correct, the discrepancy arises from the fact that the lattice trees are constructed from less branched architectures than the branched polymers having $\nu_{0} = 1/4$ in polymer physics. The present analysis suggests that the 2-dimensional lattice trees are the mixture of isomers having the mean ideal size exponent of $\bar{\nu}_{0}\doteq0.32$.
What problem does this paper attempt to address?