And why would you try to separate the xylenes? The boiling points of the three isomers are separated by just 6°C so distillation is not an option unless you carry a lot of plates. Lusi & Barbour venture ahead anyway (DOI) and draw inspiration from an 50 year article on xylene separation by clathration (DOI). Key is a so-called Werner complex based on octahedral nickel with 4 phenylpyridine and two isothiocyanate ligands. This compound forms an host-guest compound with each of the xylenes and what matters to Lusi & Barbour is how selective this process is. When a xylene vapor is exposed to the nickel compound, the ortho isomer gets caught first (o/m=34.2, o/p=40.5). With this isomer out of the way meta is then preferred over para (m/p= 12.7). Interestingly the trend in thermal stability and sorption kinetics for the isolated compounds display the opposite trend. With respect to guest-accessible volume the trend is meta < para < ortho. The next quote is noteworthy: in the selectivity trends in terms of packing interactions, any such evaluation would be merely speculative; we recognize that many subtle factors may influence selectivity and that any simplistic approach based on comparing densities, interactions, and so on would be inadequate in providing a holistic explanation for the observed trend in guest preferences.
And why again would you try to separate the xylenes? Pure meta-xylene is used as an anti-knocking agent in gasoline. Big business assured.